Split (3)

train · 30.4k rows

aid stringlengths 9 15	mid stringlengths 7 10	abstract stringlengths 78 2.56k	related_work stringlengths 92 1.77k	ref_abstract dict
1812.01748	2951781701	Modeling fashion compatibility is challenging due to its complexity and subjectivity. Existing work focuses on predicting compatibility between product images (e.g. an image containing a t-shirt and an image containing a pair of jeans). However, these approaches ignore real-world 'scene' images (e.g. selfies); such images are hard to deal with due to their complexity, clutter, variations in lighting and pose (etc.) but on the other hand could potentially provide key context (e.g. the user's body type, or the season) for making more accurate recommendations. In this work, we propose a new task called 'Complete the Look', which seeks to recommend visually compatible products based on scene images. We design an approach to extract training data for this task, and propose a novel way to learn the scene-product compatibility from fashion or interior design images. Our approach measures compatibility both globally and locally via CNNs and attention mechanisms. Extensive experiments show that our method achieves significant performance gains over alternative systems. Human evaluation and qualitative analysis are also conducted to further understand model behavior. We hope this work could lead to useful applications which link large corpora of real-world scenes with shoppable products.	Recently, computer vision for fashion has attracted significant attention, with various applications typically built on top of deep convolutional networks. Clothing parsing' is one such application, which seeks to parse and categorize garments in a fashion image @cite_6 @cite_1 @cite_10 . Since clothing has fine-graind style attributes (e.g. sleeve length, material, etc.), some works seek to identify clothing attributes @cite_20 @cite_21 @cite_16 , and detect fashion landmarks (e.g. sleeve, collar, etc.) @cite_2 @cite_33 . Another line of work considers retrieving fashion images based on various forms of queries, including images @cite_33 @cite_3 , attributes @cite_4 @cite_22 , occasions @cite_57 , videos @cite_47 , and user preferences @cite_53 . Our work is closer to the cross-scenario' fashion retrieval setting (called street2shop) which seeks to retrieve fashion products appearing in street photos @cite_12 @cite_17 , as the same type of data can be adapted to our setting.	{ "cite_N": [ "@cite_4", "@cite_33", "@cite_22", "@cite_53", "@cite_21", "@cite_1", "@cite_6", "@cite_3", "@cite_57", "@cite_2", "@cite_47", "@cite_16", "@cite_10", "@cite_12", "@cite_20", "@cite_17" ], "mid": [ "", "2471768434", "", "", "181871703", "2313077179", "2074621908", "", "2136340547", "", "2738875051", "2951538710", "2115091888", "2135367695", "2463470988", "" ], "abstract": [ "", "Recent advances in clothes recognition have been driven by the construction of clothes datasets. Existing datasets are limited in the amount of annotations and are difficult to cope with the various challenges in real-world applications. In this work, we introduce DeepFashion1, a large-scale clothes dataset with comprehensive annotations. It contains over 800,000 images, which are richly annotated with massive attributes, clothing landmarks, and correspondence of images taken under different scenarios including store, street snapshot, and consumer. Such rich annotations enable the development of powerful algorithms in clothes recognition and facilitating future researches. To demonstrate the advantages of DeepFashion, we propose a new deep model, namely FashionNet, which learns clothing features by jointly predicting clothing attributes and landmarks. The estimated landmarks are then employed to pool or gate the learned features. It is optimized in an iterative manner. Extensive experiments demonstrate the effectiveness of FashionNet and the usefulness of DeepFashion.", "", "", "The clothing we wear and our identities are closely tied, revealing to the world clues about our wealth, occupation, and socio-identity. In this paper we examine questions related to what our clothing reveals about our personal style. We first design an online competitive Style Rating Game called Hipster Wars to crowd source reliable human judgments of style. We use this game to collect a new dataset of clothing outfits with associated style ratings for 5 style categories: hipster, bohemian, pinup, preppy, and goth. Next, we train models for between-class and within-class classification of styles. Finally, we explore methods to identify clothing elements that are generally discriminative for a style, and methods for identifying items in a particular outfit that may indicate a style.", "This paper aims at developing an integrated system for clothing co-parsing (CCP), in order to jointly parse a set of clothing images (unsegmented but annotated with tags) into semantic configurations. A novel data-driven system consisting of two phases of inference is proposed. The first phase, referred as “image cosegmentation,” iterates to extract consistent regions on images and jointly refines the regions over all images by employing the exemplar-SVM technique [1] . In the second phase (i.e., “region colabeling”), we construct a multiimage graphical model by taking the segmented regions as vertices, and incorporating several contexts of clothing configuration (e.g., item locations and mutual interactions). The joint label assignment can be solved using the efficient Graph Cuts algorithm. In addition to evaluate our framework on the Fashionista dataset [2] , we construct a dataset called the SYSU-Clothes dataset consisting of 2098 high-resolution street fashion photos to demonstrate the performance of our system. We achieve 90.29 88.23 segmentation accuracy and 65.52 63.89 recognition rate on the Fashionista and the SYSU-Clothes datasets, respectively, which are superior compared with the previous methods. Furthermore, we apply our method on a challenging task, i.e., cross-domain clothing retrieval: given user photo depicting a clothing image, retrieving the same clothing items from online shopping stores based on the fine-grained parsing results.", "In this paper we demonstrate an effective method for parsing clothing in fashion photographs, an extremely challenging problem due to the large number of possible garment items, variations in configuration, garment appearance, layering, and occlusion. In addition, we provide a large novel dataset and tools for labeling garment items, to enable future research on clothing estimation. Finally, we present intriguing initial results on using clothing estimates to improve pose identification, and demonstrate a prototype application for pose-independent visual garment retrieval.", "", "In this paper, we aim at a practical system, magic closet, for automatic occasion-oriented clothing recommendation. Given a user-input occasion, e.g., wedding, shopping or dating, magic closet intelligently suggests the most suitable clothing from the user's own clothing photo album, or automatically pairs the user-specified reference clothing (upper-body or lower-body) with the most suitable one from online shops. Two key criteria are explicitly considered for the magic closet system. One criterion is to wear properly, e.g., compared to suit pants, it is more decent to wear a cocktail dress for a banquet occasion. The other criterion is to wear aesthetically, e.g., a red T-shirt matches better white pants than green pants. To narrow the semantic gap between the low-level features of clothing and the high-level occasion categories, we adopt middle-level clothing attributes (e.g., clothing category, color, pattern) as a bridge. More specifically, the clothing attributes are treated as latent variables in our proposed latent Support Vector Machine (SVM) based recommendation model. The wearing properly criterion is described in the model through a feature-occasion potential and an attribute-occasion potential, while the wearing aesthetically criterion is expressed by an attribute-attribute potential. To learn a generalize-well model and comprehensively evaluate it, we collect a large clothing What-to-Wear (WoW) dataset, and thoroughly annotate the whole dataset with 7 multi-value clothing attributes and 10 occasion categories via Amazon Mechanic Turk. Extensive experiments on the WoW dataset demonstrate the effectiveness of the magic closet system for both occasion-oriented clothing recommendation and pairing.", "", "In recent years, both online retail and video hosting service are exponentially growing. In this paper, we explore a new cross-domain task, Video2Shop, targeting for matching clothes appeared in videos to the exact same items in online shops. A novel deep neural network, called AsymNet, is proposed to explore this problem. For the image side, well- established methods are used to detect and extract features for clothing patches with arbitrary sizes. For the video side, deep visual features are extracted from detected object re- gions in each frame, and further fed into a Long Short-Term Memory (LSTM) framework for sequence modeling, which captures the temporal dynamics in videos. To conduct exact matching between videos and online shopping images, LSTM hidden states, representing the video, and image features, which represent static object images, are jointly mod- eled under the similarity network with reconfigurable deep tree structure. Moreover, an approximate training method is proposed to achieve the efficiency when training. Extensive experiments conducted on a large cross-domain dataset have demonstrated the effectiveness and efficiency of the proposed AsymNet, which outperforms the state-of-the-art methods.", "What is the future of fashion? Tackling this question from a data-driven vision perspective, we propose to forecast visual style trends before they occur. We introduce the first approach to predict the future popularity of styles discovered from fashion images in an unsupervised manner. Using these styles as a basis, we train a forecasting model to represent their trends over time. The resulting model can hypothesize new mixtures of styles that will become popular in the future, discover style dynamics (trendy vs. classic), and name the key visual attributes that will dominate tomorrow's fashion. We demonstrate our idea applied to three datasets encapsulating 80,000 fashion products sold across six years on Amazon. Results indicate that fashion forecasting benefits greatly from visual analysis, much more than textual or meta-data cues surrounding products.", "Clothing recognition is a societ ally and commercially important yet extremely challenging problem due to large variations in clothing appearance, layering, style, and body shape and pose. In this paper, we tackle the clothing parsing problem using a retrieval-based approach. For a query image, we find similar styles from a large database of tagged fashion images and use these examples to recognize clothing items in the query. Our approach combines parsing from: pre-trained global clothing models, local clothing models learned on the fly from retrieved examples, and transferred parse-masks (Paper Doll item transfer) from retrieved examples. We evaluate our approach extensively and show significant improvements over previous state-of-the-art for both localization (clothing parsing given weak supervision in the form of tags) and detection (general clothing parsing). Our experimental results also indicate that the general pose estimation problem can benefit from clothing parsing.", "We address a cross-scenario clothing retrieval problem- given a daily human photo captured in general environment, e.g., on street, finding similar clothing in online shops, where the photos are captured more professionally and with clean background. There are large discrepancies between daily photo scenario and online shopping scenario. We first propose to alleviate the human pose discrepancy by locating 30 human parts detected by a well trained human detector. Then, founded on part features, we propose a two-step calculation to obtain more reliable one-to-many similarities between the query daily photo and online shopping photos: 1) the within-scenario one-to-many similarities between a query daily photo and an extra auxiliary set are derived by direct sparse reconstruction; 2) by a cross-scenario many-to-many similarity transfer matrix inferred offline from the auxiliary set and the online shopping set, the reliable cross-scenario one-to-many similarities between the query daily photo and all online shopping photos are obtained.", "We propose a novel approach for learning features from weakly-supervised data by joint ranking and classification. In order to exploit data with weak labels, we jointly train a feature extraction network with a ranking loss and a classification network with a cross-entropy loss. We obtain high-quality compact discriminative features with few parameters, learned on relatively small datasets without additional annotations. This enables us to tackle tasks with specialized images not very similar to the more generic ones in existing fully-supervised datasets. We show that the resulting features in combination with a linear classifier surpass the state-of-the-art on the Hipster Wars dataset despite using features only 0.3 of the size. Our proposed features significantly outperform those obtained from networks trained on ImageNet, despite being 32 times smaller (128 single-precision floats), trained on noisy and weakly-labeled data, and using only 1.5 of the number of parameters.1.", "" ] }
1812.01748	2951781701	Modeling fashion compatibility is challenging due to its complexity and subjectivity. Existing work focuses on predicting compatibility between product images (e.g. an image containing a t-shirt and an image containing a pair of jeans). However, these approaches ignore real-world 'scene' images (e.g. selfies); such images are hard to deal with due to their complexity, clutter, variations in lighting and pose (etc.) but on the other hand could potentially provide key context (e.g. the user's body type, or the season) for making more accurate recommendations. In this work, we propose a new task called 'Complete the Look', which seeks to recommend visually compatible products based on scene images. We design an approach to extract training data for this task, and propose a novel way to learn the scene-product compatibility from fashion or interior design images. Our approach measures compatibility both globally and locally via CNNs and attention mechanisms. Extensive experiments show that our method achieves significant performance gains over alternative systems. Human evaluation and qualitative analysis are also conducted to further understand model behavior. We hope this work could lead to useful applications which link large corpora of real-world scenes with shoppable products.	Some recent works seek to identify whether two products are complementary, such that we can recommend complementary products based on the user's previous purchasing or browsing patterns @cite_49 @cite_0 @cite_28 . In the fashion domain, visual features can be useful to determine compatibility between items, for example in terms of pairwise compatibility @cite_29 @cite_26 @cite_58 @cite_56 , or outfit compatibility @cite_8 @cite_45 @cite_5 @cite_25 . The former setting takes a fashion item as a query and seeks to recommend compatible items from different categories (e.g. recommend jeans given a t-shirt). The latter seeks to select fashion items to form compatible outfits, or to complete a partial outfit. Our method retrieves compatible products based on a real-world scene containing rich context (e.g. garments, body shapes, occasions), which can also be viewed as a form of complementary recommendation. However this differs from existing methods which seek to model product-product compatibility from pairs of images containing products. In addition to retrieving existing products, one recent approach uses generative models to generate compatible fashion items @cite_14 .	{ "cite_N": [ "@cite_26", "@cite_14", "@cite_8", "@cite_28", "@cite_29", "@cite_56", "@cite_0", "@cite_45", "@cite_49", "@cite_5", "@cite_58", "@cite_25" ], "mid": [ "2767109396", "2774458425", "2138621090", "2782696945", "2953040449", "2027731328", "2892485145", "2737102415", "2949988325", "2949291161", "2951555918", "" ], "abstract": [ "Nowadays, as a beauty-enhancing product, clothing plays an important role in human's social life. In fact, the key to a proper outfit usually lies in the harmonious clothing matching. Nevertheless, not everyone is good at clothing matching. Fortunately, with the proliferation of fashion-oriented online communities, fashion experts can publicly share their fashion tips by showcasing their outfit compositions, where each fashion item (e.g., a top or bottom) usually has an image and context metadata (e.g., title and category). Such rich fashion data offer us a new opportunity to investigate the code in clothing matching. However, challenges co-exist with opportunities. The first challenge lies in the complicated factors, such as color, material and shape, that affect the compatibility of fashion items. Second, as each fashion item involves multiple modalities (i.e., image and text), how to cope with the heterogeneous multi-modal data also poses a great challenge. Third, our pilot study shows that the composition relation between fashion items is rather sparse, which makes traditional matrix factorization methods not applicable. Towards this end, in this work, we propose a content-based neural scheme to model the compatibility between fashion items based on the Bayesian personalized ranking (BPR) framework. The scheme is able to jointly model the coherent relation between modalities of items and their implicit matching preference. Experiments verify the effectiveness of our scheme, and we deliver deep insights that can benefit future research.", "Compatibility between items, such as clothes and shoes, is a major factor among customer's purchasing decisions. However, learning \"compatibility\" is challenging due to (1) broader notions of compatibility than those of similarity, (2) the asymmetric nature of compatibility, and (3) only a small set of compatible and incompatible items are observed. We propose an end-to-end trainable system to embed each item into a latent vector and project a query item into K compatible prototypes in the same space. These prototypes reflect the broad notions of compatibility. We refer to both the embedding and prototypes as \"Compatibility Family\". In our learned space, we introduce a novel Projected Compatibility Distance (PCD) function which is differentiable and ensures diversity by aiming for at least one prototype to be close to a compatible item, whereas none of the prototypes are close to an incompatible item. We evaluate our system on a toy dataset, two Amazon product datasets, and Polyvore outfit dataset. Our method consistently achieves state-of-the-art performance. Finally, we show that we can visualize the candidate compatible prototypes using a Metric-regularized Conditional Generative Adversarial Network (MrCGAN), where the input is a projected prototype and the output is a generated image of a compatible item. We ask human evaluators to judge the relative compatibility between our generated images and images generated by CGANs conditioned directly on query items. Our generated images are significantly preferred, with roughly twice the number of votes as others.", "Dimensionality reduction involves mapping a set of high dimensional input points onto a low dimensional manifold so that 'similar\" points in input space are mapped to nearby points on the manifold. We present a method - called Dimensionality Reduction by Learning an Invariant Mapping (DrLIM) - for learning a globally coherent nonlinear function that maps the data evenly to the output manifold. The learning relies solely on neighborhood relationships and does not require any distancemeasure in the input space. The method can learn mappings that are invariant to certain transformations of the inputs, as is demonstrated with a number of experiments. Comparisons are made to other techniques, in particular LLE.", "In personalized recommendation, candidate generation plays an infrastructural role by retrieving candidates out of billions of items. During this process, substitutes and complements constitute two main classes of retrieved candidates: substitutable products are interchangeable, whereas complementary products might be purchased together by users. Discriminating substitutable and complementary products is playing an increasingly important role in e-commerce portals by affecting the performance of candidate generation, e.g., when a user has browsed a t-shirt, it is reasonable to retrieve similar t-shirts, i.e., substitutes; whereas if the user has already purchased one, it would be better to retrieve trousers, hats or shoes, as complements of t-shirts. In this paper, we propose a path-constrained framework (PMSC) for discriminating substitutes and complements. Specifically, for each product, we first learn its embedding representations in a general semantic space. Thereafter, we project the embedding vectors into two separate spaces via a novel mapping function. In the end, we incorporate each embedding with path-constraints to further boost the discriminative ability of the model. Extensive experiments conducted on two e-commerce datasets show the effectiveness of our proposed method.", "With the rapid proliferation of smart mobile devices, users now take millions of photos every day. These include large numbers of clothing and accessory images. We would like to answer questions like What outfit goes well with this pair of shoes?' To answer these types of questions, one has to go beyond learning visual similarity and learn a visual notion of compatibility across categories. In this paper, we propose a novel learning framework to help answer these types of questions. The main idea of this framework is to learn a feature transformation from images of items into a latent space that expresses compatibility. For the feature transformation, we use a Siamese Convolutional Neural Network (CNN) architecture, where training examples are pairs of items that are either compatible or incompatible. We model compatibility based on co-occurrence in large-scale user behavior data; in particular co-purchase data from Amazon.com. To learn cross-category fit, we introduce a strategic method to sample training data, where pairs of items are heterogeneous dyads, i.e., the two elements of a pair belong to different high-level categories. While this approach is applicable to a wide variety of settings, we focus on the representative problem of learning compatible clothing style. Our results indicate that the proposed framework is capable of learning semantic information about visual style and is able to generate outfits of clothes, with items from different categories, that go well together.", "Humans inevitably develop a sense of the relationships between objects, some of which are based on their appearance. Some pairs of objects might be seen as being alternatives to each other (such as two pairs of jeans), while others may be seen as being complementary (such as a pair of jeans and a matching shirt). This information guides many of the choices that people make, from buying clothes to their interactions with each other. We seek here to model this human sense of the relationships between objects based on their appearance. Our approach is not based on fine-grained modeling of user annotations but rather on capturing the largest dataset possible and developing a scalable method for uncovering human notions of the visual relationships within. We cast this as a network inference problem defined on graphs of related images, and provide a large-scale dataset for the training and evaluation of the same. The system we develop is capable of recommending which clothes and accessories will go well together (and which will not), amongst a host of other applications.", "Complementary item recommendation finds products that go well with one another (e.g., a camera and a specific lens). While complementary items are ubiquitous, the dimensions by which items go together can vary by both product and category, making it difficult to detect complementary items at scale. Moreover, in practice, user preferences for complementary items can be complex combinations of item quality and evidence of complementarity. Hence, we propose a new neural complementary recommender E ncore that can jointly learn complementary item relationships and user preferences. Specifically, E ncore (i) effectively combines and balances both stylistic and functional evidence of complementary items across item categories; (ii) naturally models item latent quality for complementary items through Bayesian inference of customer ratings; and (iii) builds a novel neural network model to learn the complex (non-linear) relationships between items for flexible and scalable complementary product recommendations. Through experiments over large Amazon datasets, we find that E ncore effectively learns complementary item relationships, leading to an improvement in accuracy of 15.5 on average versus the next-best alternative.", "The ubiquity of online fashion shopping demands effective recommendation services for customers. In this paper, we study two types of fashion recommendation: (i) suggesting an item that matches existing components in a set to form a stylish outfit (a collection of fashion items), and (ii) generating an outfit with multimodal (images text) specifications from a user. To this end, we propose to jointly learn a visual-semantic embedding and the compatibility relationships among fashion items in an end-to-end fashion. More specifically, we consider a fashion outfit to be a sequence (usually from top to bottom and then accessories) and each item in the outfit as a time step. Given the fashion items in an outfit, we train a bidirectional LSTM (Bi-LSTM) model to sequentially predict the next item conditioned on previous ones to learn their compatibility relationships. Further, we learn a visual-semantic space by regressing image features to their semantic representations aiming to inject attribute and category information as a regularization for training the LSTM. The trained network can not only perform the aforementioned recommendations effectively but also predict the compatibility of a given outfit. We conduct extensive experiments on our newly collected Polyvore dataset, and the results provide strong qualitative and quantitative evidence that our framework outperforms alternative methods.", "In a modern recommender system, it is important to understand how products relate to each other. For example, while a user is looking for mobile phones, it might make sense to recommend other phones, but once they buy a phone, we might instead want to recommend batteries, cases, or chargers. These two types of recommendations are referred to as substitutes and complements: substitutes are products that can be purchased instead of each other, while complements are products that can be purchased in addition to each other. Here we develop a method to infer networks of substitutable and complementary products. We formulate this as a supervised link prediction task, where we learn the semantics of substitutes and complements from data associated with products. The primary source of data we use is the text of product reviews, though our method also makes use of features such as ratings, specifications, prices, and brands. Methodologically, we build topic models that are trained to automatically discover topics from text that are successful at predicting and explaining such relationships. Experimentally, we evaluate our system on the Amazon product catalog, a large dataset consisting of 9 million products, 237 million links, and 144 million reviews.", "We propose to automatically create capsule wardrobes. Given an inventory of candidate garments and accessories, the algorithm must assemble a minimal set of items that provides maximal mix-and-match outfits. We pose the task as a subset selection problem. To permit efficient subset selection over the space of all outfit combinations, we develop submodular objective functions capturing the key ingredients of visual compatibility, versatility, and user-specific preference. Since adding garments to a capsule only expands its possible outfits, we devise an iterative approach to allow near-optimal submodular function maximization. Finally, we present an unsupervised approach to learn visual compatibility from \"in the wild\" full body outfit photos; the compatibility metric translates well to cleaner catalog photos and improves over existing methods. Our results on thousands of pieces from popular fashion websites show that automatic capsule creation has potential to mimic skilled fashionistas in assembling flexible wardrobes, while being significantly more scalable.", "Outfits in online fashion data are composed of items of many different types (e.g. top, bottom, shoes) that share some stylistic relationship with one another. A representation for building outfits requires a method that can learn both notions of similarity (for example, when two tops are interchangeable) and compatibility (items of possibly different type that can go together in an outfit). This paper presents an approach to learning an image embedding that respects item type, and jointly learns notions of item similarity and compatibility in an end-to-end model. To evaluate the learned representation, we crawled 68,306 outfits created by users on the Polyvore website. Our approach obtains 3-5 improvement over the state-of-the-art on outfit compatibility prediction and fill-in-the-blank tasks using our dataset, as well as an established smaller dataset, while supporting a variety of useful queries.", "" ] }
1812.01748	2951781701	Modeling fashion compatibility is challenging due to its complexity and subjectivity. Existing work focuses on predicting compatibility between product images (e.g. an image containing a t-shirt and an image containing a pair of jeans). However, these approaches ignore real-world 'scene' images (e.g. selfies); such images are hard to deal with due to their complexity, clutter, variations in lighting and pose (etc.) but on the other hand could potentially provide key context (e.g. the user's body type, or the season) for making more accurate recommendations. In this work, we propose a new task called 'Complete the Look', which seeks to recommend visually compatible products based on scene images. We design an approach to extract training data for this task, and propose a novel way to learn the scene-product compatibility from fashion or interior design images. Our approach measures compatibility both globally and locally via CNNs and attention mechanisms. Extensive experiments show that our method achieves significant performance gains over alternative systems. Human evaluation and qualitative analysis are also conducted to further understand model behavior. We hope this work could lead to useful applications which link large corpora of real-world scenes with shoppable products.	Attention' has been widely used in computer vision tasks including image captioning @cite_37 @cite_41 , visual question answering @cite_50 @cite_13 , image recognition @cite_55 @cite_48 , and generation @cite_36 @cite_34 . Attention is mainly used to focus' on relevant regions of an image (known as spatial attention'). To identify relevant regions of fashion images, previous methods have adopted pretrained person detectors to segment images @cite_12 @cite_23 . Another approach discovers relevant regions by attribute activation maps (AAMs) @cite_54 , generated using labels including clothing attributes @cite_22 and descriptions @cite_44 . Recently, attention mechanisms have achieved strong performance on visual fashion understanding tasks like clothing categorization and fashion landmark detection @cite_2 . Our work is the first (to our knowledge) to apply attention to discover relevant regions guided by supervision in the form of compatibility.	{ "cite_N": [ "@cite_37", "@cite_22", "@cite_41", "@cite_48", "@cite_55", "@cite_36", "@cite_54", "@cite_2", "@cite_44", "@cite_50", "@cite_23", "@cite_34", "@cite_13", "@cite_12" ], "mid": [ "2950178297", "", "2550553598", "", "2951005624", "2771088323", "2950328304", "", "2742448362", "2179022885", "2155171143", "2950893734", "2255577267", "2135367695" ], "abstract": [ "Inspired by recent work in machine translation and object detection, we introduce an attention based model that automatically learns to describe the content of images. We describe how we can train this model in a deterministic manner using standard backpropagation techniques and stochastically by maximizing a variational lower bound. We also show through visualization how the model is able to automatically learn to fix its gaze on salient objects while generating the corresponding words in the output sequence. We validate the use of attention with state-of-the-art performance on three benchmark datasets: Flickr8k, Flickr30k and MS COCO.", "", "Visual attention has been successfully applied in structural prediction tasks such as visual captioning and question answering. Existing visual attention models are generally spatial, i.e., the attention is modeled as spatial probabilities that re-weight the last conv-layer feature map of a CNN encoding an input image. However, we argue that such spatial attention does not necessarily conform to the attention mechanism — a dynamic feature extractor that combines contextual fixations over time, as CNN features are naturally spatial, channel-wise and multi-layer. In this paper, we introduce a novel convolutional neural network dubbed SCA-CNN that incorporates Spatial and Channel-wise Attentions in a CNN. In the task of image captioning, SCA-CNN dynamically modulates the sentence generation context in multi-layer feature maps, encoding where (i.e., attentive spatial locations at multiple layers) and what (i.e., attentive channels) the visual attention is. We evaluate the proposed SCA-CNN architecture on three benchmark image captioning datasets: Flickr8K, Flickr30K, and MSCOCO. It is consistently observed that SCA-CNN significantly outperforms state-of-the-art visual attention-based image captioning methods.", "", "Convolutional Neural Networks define an exceptionally powerful class of models, but are still limited by the lack of ability to be spatially invariant to the input data in a computationally and parameter efficient manner. In this work we introduce a new learnable module, the Spatial Transformer, which explicitly allows the spatial manipulation of data within the network. This differentiable module can be inserted into existing convolutional architectures, giving neural networks the ability to actively spatially transform feature maps, conditional on the feature map itself, without any extra training supervision or modification to the optimisation process. We show that the use of spatial transformers results in models which learn invariance to translation, scale, rotation and more generic warping, resulting in state-of-the-art performance on several benchmarks, and for a number of classes of transformations.", "In this paper, we propose an Attentional Generative Adversarial Network (AttnGAN) that allows attention-driven, multi-stage refinement for fine-grained text-to-image generation. With a novel attentional generative network, the AttnGAN can synthesize fine-grained details at different subregions of the image by paying attentions to the relevant words in the natural language description. In addition, a deep attentional multimodal similarity model is proposed to compute a fine-grained image-text matching loss for training the generator. The proposed AttnGAN significantly outperforms the previous state of the art, boosting the best reported inception score by 14.14 on the CUB dataset and 170.25 on the more challenging COCO dataset. A detailed analysis is also performed by visualizing the attention layers of the AttnGAN. It for the first time shows that the layered attentional GAN is able to automatically select the condition at the word level for generating different parts of the image.", "In this work, we revisit the global average pooling layer proposed in [13], and shed light on how it explicitly enables the convolutional neural network to have remarkable localization ability despite being trained on image-level labels. While this technique was previously proposed as a means for regularizing training, we find that it actually builds a generic localizable deep representation that can be applied to a variety of tasks. Despite the apparent simplicity of global average pooling, we are able to achieve 37.1 top-5 error for object localization on ILSVRC 2014, which is remarkably close to the 34.2 top-5 error achieved by a fully supervised CNN approach. We demonstrate that our network is able to localize the discriminative image regions on a variety of tasks despite not being trained for them", "", "This paper proposes an automatic spatially-aware concept discovery approach using weakly labeled image-text data from shopping websites. We first fine-tune GoogleNet by jointly modeling clothing images and their corresponding descriptions in a visual-semantic embedding space. Then, for each attribute (word), we generate its spatially-aware representation by combining its semantic word vector representation with its spatial representation derived from the convolutional maps of the fine-tuned network. The resulting spatially-aware representations are further used to cluster attributes into multiple groups to form spatially-aware concepts (e.g., the neckline concept might consist of attributes like v-neck, round-neck, etc). Finally, we decompose the visual-semantic embedding space into multiple concept-specific subspaces, which facilitates structured browsing and attribute-feedback product retrieval by exploiting multimodal linguistic regularities. We conducted extensive experiments on our newly collected Fashion200K dataset, and results on clustering quality evaluation and attribute-feedback product retrieval task demonstrate the effectiveness of our automatically discovered spatially-aware concepts.", "We present a method that learns to answer visual questions by selecting image regions relevant to the text-based query. Our method exhibits significant improvements in answering questions such as \"what color,\" where it is necessary to evaluate a specific location, and \"what room,\" where it selectively identifies informative image regions. Our model is tested on the VQA dataset which is the largest human-annotated visual question answering dataset to our knowledge.", "Predicting human occupations in photos has great application potentials in intelligent services and systems. However, using traditional classification methods cannot reliably distinguish different occupations due to the complex relations between occupations and the low-level image features. In this paper, we investigate the human occupation prediction problem by modeling the appearances of human clothing as well as surrounding context. The human clothing, regarding its complex details and variant appearances, is described via part-based modeling on the automatically aligned patches of human body parts. The image patches are represented with semantic-level patterns such as clothes and haircut styles using methods based on sparse coding towards informative and noise-tolerant capacities. This description of human clothing is proved to be more effective than traditional methods. Different kinds of surrounding context are also investigated as a complementarity of human clothing features in the cases that the background information is available. Experiments are conducted on a well labeled image database that contains more than 5; 000 images from 20 representative occupation categories. The preliminary study shows the human occupation is reasonably predictable using the proposed clothing features and possible context.", "In this paper, we propose the Self-Attention Generative Adversarial Network (SAGAN) which allows attention-driven, long-range dependency modeling for image generation tasks. Traditional convolutional GANs generate high-resolution details as a function of only spatially local points in lower-resolution feature maps. In SAGAN, details can be generated using cues from all feature locations. Moreover, the discriminator can check that highly detailed features in distant portions of the image are consistent with each other. Furthermore, recent work has shown that generator conditioning affects GAN performance. Leveraging this insight, we apply spectral normalization to the GAN generator and find that this improves training dynamics. The proposed SAGAN achieves the state-of-the-art results, boosting the best published Inception score from 36.8 to 52.52 and reducing Frechet Inception distance from 27.62 to 18.65 on the challenging ImageNet dataset. Visualization of the attention layers shows that the generator leverages neighborhoods that correspond to object shapes rather than local regions of fixed shape.", "We address the problem of Visual Question Answering (VQA), which requires joint image and language understanding to answer a question about a given photograph. Recent approaches have applied deep image captioning methods based on convolutional-recurrent networks to this problem, but have failed to model spatial inference. To remedy this, we propose a model we call the Spatial Memory Network and apply it to the VQA task. Memory networks are recurrent neural networks with an explicit attention mechanism that selects certain parts of the information stored in memory. Our Spatial Memory Network stores neuron activations from different spatial regions of the image in its memory, and uses the question to choose relevant regions for computing the answer, a process of which constitutes a single \"hop\" in the network. We propose a novel spatial attention architecture that aligns words with image patches in the first hop, and obtain improved results by adding a second attention hop which considers the whole question to choose visual evidence based on the results of the first hop. To better understand the inference process learned by the network, we design synthetic questions that specifically require spatial inference and visualize the attention weights. We evaluate our model on two published visual question answering datasets, DAQUAR [1] and VQA [2], and obtain improved results compared to a strong deep baseline model (iBOWIMG) which concatenates image and question features to predict the answer [3].", "We address a cross-scenario clothing retrieval problem- given a daily human photo captured in general environment, e.g., on street, finding similar clothing in online shops, where the photos are captured more professionally and with clean background. There are large discrepancies between daily photo scenario and online shopping scenario. We first propose to alleviate the human pose discrepancy by locating 30 human parts detected by a well trained human detector. Then, founded on part features, we propose a two-step calculation to obtain more reliable one-to-many similarities between the query daily photo and online shopping photos: 1) the within-scenario one-to-many similarities between a query daily photo and an extra auxiliary set are derived by direct sparse reconstruction; 2) by a cross-scenario many-to-many similarity transfer matrix inferred offline from the auxiliary set and the online shopping set, the reliable cross-scenario one-to-many similarities between the query daily photo and all online shopping photos are obtained." ] }
1812.01640	2902875180	Catastrophic forgetting is a challenge issue in continual learning when a deep neural network forgets the knowledge acquired from the former task after learning on subsequent tasks. However, existing methods try to find the joint distribution of parameters shared with all tasks. This idea can be questionable because this joint distribution may not present when the number of tasks increase. On the other hand, It also leads to "long-term" memory issue when the network capacity is limited since adding tasks will "eat" the network capacity. In this paper, we proposed a Soft Parameters Pruning (SPP) strategy to reach the trade-off between short-term and long-term profit of a learning model by freeing those parameters less contributing to remember former task domain knowledge to learn future tasks, and preserving memories about previous tasks via those parameters effectively encoding knowledge about tasks at the same time. The SPP also measures the importance of parameters by information entropy in a label free manner. The experiments on several tasks shows SPP model achieved the best performance compared with others state-of-the-art methods. Experiment results also indicate that our method is less sensitive to hyper-parameter and better generalization. Our research suggests that a softer strategy, i.e. approximate optimize or sub-optimal solution, will benefit alleviating the dilemma of memory. The source codes are available at https: github.com lehaifeng Learning_by_memory.	Parameter pruning methods @cite_29 @cite_22 are based on the hypothesis that nonessential parameters have little effect on the model’s error after being erased and thus the key point is to search the optimum parameters that can minimum the interference to error . An effective way to narrow the representational overlap between tasks is to lessen coding parameters of representation in the continual learning model in limited capacity. Knowledge distillation pack the knowledge of complex network into a lightweight target network by the mode of teacher-student, and it also be used to tackle the problem of catastrophic forgetting.	{ "cite_N": [ "@cite_29", "@cite_22" ], "mid": [ "2114766824", "2125389748" ], "abstract": [ "We have used information-theoretic ideas to derive a class of practical and nearly optimal schemes for adapting the size of a neural network. By removing unimportant weights from a network, several improvements can be expected: better generalization, fewer training examples required, and improved speed of learning and or classification. The basic idea is to use second-derivative information to make a tradeoff between network complexity and training set error. Experiments confirm the usefulness of the methods on a real-world application.", "We investigate the use of information from all second order derivatives of the error function to perform network pruning (i.e., removing unimportant weights from a trained network) in order to improve generalization, simplify networks, reduce hardware or storage requirements, increase the speed of further training, and in some cases enable rule extraction. Our method, Optimal Brain Surgeon (OBS), is Significantly better than magnitude-based methods and Optimal Brain Damage [Le Cun, Denker and Solla, 1990], which often remove the wrong weights. OBS permits the pruning of more weights than other methods (for the same error on the training set), and thus yields better generalization on test data. Crucial to OBS is a recursion relation for calculating the inverse Hessian matrix H-1 from training data and structural information of the net. OBS permits a 90 , a 76 , and a 62 reduction in weights over backpropagation with weight decay on three benchmark MONK's problems [, 1991]. Of OBS, Optimal Brain Damage, and magnitude-based methods, only OBS deletes the correct weights from a trained XOR network in every case. Finally, whereas Sejnowski and Rosenberg [1987] used 18,000 weights in their NETtalk network, we used OBS to prune a network to just 1560 weights, yielding better generalization." ] }
1812.01640	2902875180	Catastrophic forgetting is a challenge issue in continual learning when a deep neural network forgets the knowledge acquired from the former task after learning on subsequent tasks. However, existing methods try to find the joint distribution of parameters shared with all tasks. This idea can be questionable because this joint distribution may not present when the number of tasks increase. On the other hand, It also leads to "long-term" memory issue when the network capacity is limited since adding tasks will "eat" the network capacity. In this paper, we proposed a Soft Parameters Pruning (SPP) strategy to reach the trade-off between short-term and long-term profit of a learning model by freeing those parameters less contributing to remember former task domain knowledge to learn future tasks, and preserving memories about previous tasks via those parameters effectively encoding knowledge about tasks at the same time. The SPP also measures the importance of parameters by information entropy in a label free manner. The experiments on several tasks shows SPP model achieved the best performance compared with others state-of-the-art methods. Experiment results also indicate that our method is less sensitive to hyper-parameter and better generalization. Our research suggests that a softer strategy, i.e. approximate optimize or sub-optimal solution, will benefit alleviating the dilemma of memory. The source codes are available at https: github.com lehaifeng Learning_by_memory.	PackNet @cite_26 sequentially compress multiple tasks into single model by pruning redundancy parameters to overcome catastrophic forgetting. Dual memory network drew on this idea partially to overcome catastrophic forgetting by an external network. Inspired by the idea of model compression, our method utilizes parameter-importance to set up a soft mask rather than hard pruning based on binary mask, it dose not completely truncate the unimportant parameters, but adaptive adjust to later tasks to some extent, shares part of parameters among multiple tasks and save model capacity compared to hard pruning as well as enjoy lowered performance penalties.	{ "cite_N": [ "@cite_26" ], "mid": [ "2770882886" ], "abstract": [ "This paper presents a method for adding multiple tasks to a single deep neural network while avoiding catastrophic forgetting. Inspired by network pruning techniques, we exploit redundancies in large deep networks to free up parameters that can then be employed to learn new tasks. By performing iterative pruning and network re-training, we are able to sequentially \"pack\" multiple tasks into a single network while ensuring minimal drop in performance and minimal storage overhead. Unlike prior work that uses proxy losses to maintain accuracy on older tasks, we always optimize for the task at hand. We perform extensive experiments on a variety of network architectures and large-scale datasets, and observe much better robustness against catastrophic forgetting than prior work. In particular, we are able to add three fine-grained classification tasks to a single ImageNet-trained VGG-16 network and achieve accuracies close to those of separately trained networks for each task." ] }
1812.01640	2902875180	Catastrophic forgetting is a challenge issue in continual learning when a deep neural network forgets the knowledge acquired from the former task after learning on subsequent tasks. However, existing methods try to find the joint distribution of parameters shared with all tasks. This idea can be questionable because this joint distribution may not present when the number of tasks increase. On the other hand, It also leads to "long-term" memory issue when the network capacity is limited since adding tasks will "eat" the network capacity. In this paper, we proposed a Soft Parameters Pruning (SPP) strategy to reach the trade-off between short-term and long-term profit of a learning model by freeing those parameters less contributing to remember former task domain knowledge to learn future tasks, and preserving memories about previous tasks via those parameters effectively encoding knowledge about tasks at the same time. The SPP also measures the importance of parameters by information entropy in a label free manner. The experiments on several tasks shows SPP model achieved the best performance compared with others state-of-the-art methods. Experiment results also indicate that our method is less sensitive to hyper-parameter and better generalization. Our research suggests that a softer strategy, i.e. approximate optimize or sub-optimal solution, will benefit alleviating the dilemma of memory. The source codes are available at https: github.com lehaifeng Learning_by_memory.	Weight Freezing, enlightened by distributed encoding of human brain neurons, tries to avoid overlaps between crucial functional modules of tasks. For instance, Path-Net @cite_12 sets up a huge neural network, then fixes specific function module of network to avoid being interfered by later tasks. Progressive Neural Network (PNN) @cite_20 allocates separate networks for each task and performs multitasks by progressive expansion strategy. This kind of methods fix important parameters of a task durably to prevent network from forgetting acquired knowledge. However, those methods suffer network capacity exploding from long-term tasks.	{ "cite_N": [ "@cite_20", "@cite_12" ], "mid": [ "2426267443", "2583761661" ], "abstract": [ "Methods and systems for performing a sequence of machine learning tasks. One system includes a sequence of deep neural networks (DNNs), including: a first DNN corresponding to a first machine learning task, wherein the first DNN comprises a first plurality of indexed layers, and each layer in the first plurality of indexed layers is configured to receive a respective layer input and process the layer input to generate a respective layer output; and one or more subsequent DNNs corresponding to one or more respective machine learning tasks, wherein each subsequent DNN comprises a respective plurality of indexed layers, and each layer in a respective plurality of indexed layers with index greater than one receives input from a preceding layer of the respective subsequent DNN, and one or more preceding layers of respective preceding DNNs, wherein a preceding layer is a layer whose index is one less than the current index.", "For artificial general intelligence (AGI) it would be efficient if multiple users trained the same giant neural network, permitting parameter reuse, without catastrophic forgetting. PathNet is a first step in this direction. It is a neural network algorithm that uses agents embedded in the neural network whose task is to discover which parts of the network to re-use for new tasks. Agents are pathways (views) through the network which determine the subset of parameters that are used and updated by the forwards and backwards passes of the backpropogation algorithm. During learning, a tournament selection genetic algorithm is used to select pathways through the neural network for replication and mutation. Pathway fitness is the performance of that pathway measured according to a cost function. We demonstrate successful transfer learning; fixing the parameters along a path learned on task A and re-evolving a new population of paths for task B, allows task B to be learned faster than it could be learned from scratch or after fine-tuning. Paths evolved on task B re-use parts of the optimal path evolved on task A. Positive transfer was demonstrated for binary MNIST, CIFAR, and SVHN supervised learning classification tasks, and a set of Atari and Labyrinth reinforcement learning tasks, suggesting PathNets have general applicability for neural network training. Finally, PathNet also significantly improves the robustness to hyperparameter choices of a parallel asynchronous reinforcement learning algorithm (A3C)." ] }
1812.01640	2902875180	Catastrophic forgetting is a challenge issue in continual learning when a deep neural network forgets the knowledge acquired from the former task after learning on subsequent tasks. However, existing methods try to find the joint distribution of parameters shared with all tasks. This idea can be questionable because this joint distribution may not present when the number of tasks increase. On the other hand, It also leads to "long-term" memory issue when the network capacity is limited since adding tasks will "eat" the network capacity. In this paper, we proposed a Soft Parameters Pruning (SPP) strategy to reach the trade-off between short-term and long-term profit of a learning model by freeing those parameters less contributing to remember former task domain knowledge to learn future tasks, and preserving memories about previous tasks via those parameters effectively encoding knowledge about tasks at the same time. The SPP also measures the importance of parameters by information entropy in a label free manner. The experiments on several tasks shows SPP model achieved the best performance compared with others state-of-the-art methods. Experiment results also indicate that our method is less sensitive to hyper-parameter and better generalization. Our research suggests that a softer strategy, i.e. approximate optimize or sub-optimal solution, will benefit alleviating the dilemma of memory. The source codes are available at https: github.com lehaifeng Learning_by_memory.	A classic weight consolidation method @cite_8 @cite_28 is elastic weight consolidation (EWC) @cite_3 . EWC, inspired by the mechanism of synaptic plasticity, updates parameters elastically via determining important parameters. This type of method encode more tasks knowledge with less network capacity and lower computation complexity compared with Path-Net and PNN. The upper bound of tasks EWC can learn is constrained by capacity of network which is determined by the model structure. Since the model structure is invariant during learning process, the increased tasks potentially lead to performance of degenerate of EWC.	{ "cite_N": [ "@cite_28", "@cite_3", "@cite_8" ], "mid": [ "2949268663", "", "2768412495" ], "abstract": [ "While deep learning has led to remarkable advances across diverse applications, it struggles in domains where the data distribution changes over the course of learning. In stark contrast, biological neural networks continually adapt to changing domains, possibly by leveraging complex molecular machinery to solve many tasks simultaneously. In this study, we introduce intelligent synapses that bring some of this biological complexity into artificial neural networks. Each synapse accumulates task relevant information over time, and exploits this information to rapidly store new memories without forgetting old ones. We evaluate our approach on continual learning of classification tasks, and show that it dramatically reduces forgetting while maintaining computational efficiency.", "", "Humans can learn in a continuous manner. Old rarely utilized knowledge can be overwritten by new incoming information while important, frequently used knowledge is prevented from being erased. In artificial learning systems, lifelong learning so far has focused mainly on accumulating knowledge over tasks and overcoming catastrophic forgetting. In this paper, we argue that, given the limited model capacity and the unlimited new information to be learned, knowledge has to be preserved or erased selectively. Inspired by neuroplasticity, we propose an online method to compute the importance of the parameters of a neural network, based on the data that the network is actively applied to, in an unsupervised manner. After learning a task, whenever a sample is fed to the network, we accumulate an importance measure for each parameter of the network, based on how sensitive the predicted output is to a change in this parameter. When learning a new task, changes to important parameters are penalized. We show that a local version of our method is a direct application of Hebb's rule in identifying the important connections between neurons. We test our method on a sequence of object recognition tasks and on the challenging problem of learning an embedding in a continuous manner. We show state of the art performance and the ability to adapt the importance of the parameters towards what the network needs (not) to forget, which may be different for different test conditions." ] }
1812.01821	2902448979	Despite the considerable success of convolutional neural networks in a broad array of domains, recent research has shown these to be vulnerable to small adversarial perturbations, commonly known as adversarial examples. Moreover, such examples have shown to be remarkably portable, or transferable, from one model to another, enabling highly successful black-box attacks. We explore this issue of transferability and robustness from two dimensions: first, considering the impact of conventional @math regularization as well as replacing the top layer with a linear support vector machine (SVM), and second, the value of combining regularized models into an ensemble. We show that models trained with different regularizers present barriers to transferability, as does partial information about the models comprising the ensemble.	In response to various attack methods, researchers have proposed many defensive methods against AEs targeting at neural network models @cite_1 @cite_7 @cite_0 @cite_4 . However, our goal is to analyze how transferability is influenced by the nature of normal CNNs. Thus in our experiments, we study the transferability of AEs on models that are not augmented with any defensive methods.	{ "cite_N": [ "@cite_0", "@cite_4", "@cite_1", "@cite_7" ], "mid": [ "2274565976", "2610190180", "1883420340", "2217248474" ], "abstract": [ "Advances in deep learning have led to the broad adoption of Deep Neural Networks (DNNs) to a range of important machine learning problems, e.g., guiding autonomous vehicles, speech recognition, malware detection. Yet, machine learning models, including DNNs, were shown to be vulnerable to adversarial samples-subtly (and often humanly indistinguishably) modified malicious inputs crafted to compromise the integrity of their outputs. Adversarial examples thus enable adversaries to manipulate system behaviors. Potential attacks include attempts to control the behavior of vehicles, have spam content identified as legitimate content, or have malware identified as legitimate software. Adversarial examples are known to transfer from one model to another, even if the second model has a different architecture or was trained on a different set. We introduce the first practical demonstration that this cross-model transfer phenomenon enables attackers to control a remotely hosted DNN with no access to the model, its parameters, or its training data. In our demonstration, we only assume that the adversary can observe outputs from the target DNN given inputs chosen by the adversary. We introduce the attack strategy of fitting a substitute model to the input-output pairs in this manner, then crafting adversarial examples based on this auxiliary model. We evaluate the approach on existing DNN datasets and real-world settings. In one experiment, we force a DNN supported by MetaMind (one of the online APIs for DNN classifiers) to mis-classify inputs at a rate of 84.24 . We conclude with experiments exploring why adversarial samples transfer between DNNs, and a discussion on the applicability of our attack when targeting machine learning algorithms distinct from DNNs.", "We introduce Parseval networks, a form of deep neural networks in which the Lipschitz constant of linear, convolutional and aggregation layers is constrained to be smaller than 1. Parseval networks are empirically and theoretically motivated by an analysis of the robustness of the predictions made by deep neural networks when their input is subject to an adversarial perturbation. The most important feature of Parseval networks is to maintain weight matrices of linear and convolutional layers to be (approximately) Parseval tight frames, which are extensions of orthogonal matrices to non-square matrices. We describe how these constraints can be maintained efficiently during SGD. We show that Parseval networks match the state-of-the-art in terms of accuracy on CIFAR-10 100 and Street View House Numbers (SVHN) while being more robust than their vanilla counterpart against adversarial examples. Incidentally, Parseval networks also tend to train faster and make a better usage of the full capacity of the networks.", "Recent work has shown deep neural networks (DNNs) to be highly susceptible to well-designed, small perturbations at the input layer, or so-called adversarial examples. Taking images as an example, such distortions are often imperceptible, but can result in 100 mis-classification for a state of the art DNN. We study the structure of adversarial examples and explore network topology, pre-processing and training strategies to improve the robustness of DNNs. We perform various experiments to assess the removability of adversarial examples by corrupting with additional noise and pre-processing with denoising autoencoders (DAEs). We find that DAEs can remove substantial amounts of the adversarial noise. How- ever, when stacking the DAE with the original DNN, the resulting network can again be attacked by new adversarial examples with even smaller distortion. As a solution, we propose Deep Contractive Network, a model with a new end-to-end training procedure that includes a smoothness penalty inspired by the contractive autoencoder (CAE). This increases the network robustness to adversarial examples, without a significant performance penalty.", "We show that adversarial examples, i.e., the visually imperceptible perturbations that result in Convolutional Neural Networks (CNNs) fail, can be alleviated with a mechanism based on foveations---applying the CNN in different image regions. To see this, first, we report results in ImageNet that lead to a revision of the hypothesis that adversarial perturbations are a consequence of CNNs acting as a linear classifier: CNNs act locally linearly to changes in the image regions with objects recognized by the CNN, and in other regions the CNN may act non-linearly. Then, we corroborate that when the neural responses are linear, applying the foveation mechanism to the adversarial example tends to significantly reduce the effect of the perturbation. This is because, hypothetically, the CNNs for ImageNet are robust to changes of scale and translation of the object produced by the foveation, but this property does not generalize to transformations of the perturbation. As a result, the accuracy after a foveation is almost the same as the accuracy of the CNN without the adversarial perturbation, even if the adversarial perturbation is calculated taking into account a foveation." ] }
1812.01821	2902448979	Despite the considerable success of convolutional neural networks in a broad array of domains, recent research has shown these to be vulnerable to small adversarial perturbations, commonly known as adversarial examples. Moreover, such examples have shown to be remarkably portable, or transferable, from one model to another, enabling highly successful black-box attacks. We explore this issue of transferability and robustness from two dimensions: first, considering the impact of conventional @math regularization as well as replacing the top layer with a linear support vector machine (SVM), and second, the value of combining regularized models into an ensemble. We show that models trained with different regularizers present barriers to transferability, as does partial information about the models comprising the ensemble.	Based on researches on transferability, @cite_10 manages to develop a new black-box attacks methods which are based on ensemble models to create AEs with higher transferability. Our research instead analyzes how to generate AEs with high or low transferability using existing attack methods by manipulating target models.	{ "cite_N": [ "@cite_10" ], "mid": [ "2950864148" ], "abstract": [ "An intriguing property of deep neural networks is the existence of adversarial examples, which can transfer among different architectures. These transferable adversarial examples may severely hinder deep neural network-based applications. Previous works mostly study the transferability using small scale datasets. In this work, we are the first to conduct an extensive study of the transferability over large models and a large scale dataset, and we are also the first to study the transferability of targeted adversarial examples with their target labels. We study both non-targeted and targeted adversarial examples, and show that while transferable non-targeted adversarial examples are easy to find, targeted adversarial examples generated using existing approaches almost never transfer with their target labels. Therefore, we propose novel ensemble-based approaches to generating transferable adversarial examples. Using such approaches, we observe a large proportion of targeted adversarial examples that are able to transfer with their target labels for the first time. We also present some geometric studies to help understanding the transferable adversarial examples. Finally, we show that the adversarial examples generated using ensemble-based approaches can successfully attack this http URL, which is a black-box image classification system." ] }
1812.01735	2950748128	We consider the problem of efficiently constructing cheap and novel round trip flight itineraries by combining legs from different airlines. We analyse the factors that contribute towards the price of such itineraries and find that many result from the combination of just 30 of airlines and that the closer the departure of such itineraries is to the user's search date the more likely they are to be cheaper than the tickets from one airline. We use these insights to formulate the problem as a trade-off between the recall of cheap itinerary constructions and the costs associated with building them. We propose a supervised learning solution with location embeddings which achieves an AUC=80.48, a substantial improvement over simpler baselines. We discuss various practical considerations for dealing with the staleness and the stability of the model and present the design of the machine learning pipeline. Finally, we present an analysis of the model's performance in production and its impact on Skyscanner's users.	The problem of airline fare prediction is discussed in detail in @cite_16 and several data mining models were benchmarked in @cite_10 . The authors of @cite_19 modelled 3D trajectories of flights based on various weather and air traffic conditions. The problem of itinerary relevance ranking in one of the largest Global Distributed Systems was presented in @cite_18 . The systematic patterns of airline delays were analysed in @cite_4 . And the impact of airport network structure on the spread of global pandemics was weighed up in @cite_15 .	{ "cite_N": [ "@cite_18", "@cite_4", "@cite_19", "@cite_15", "@cite_16", "@cite_10" ], "mid": [ "2743586218", "", "2511826256", "2043890595", "594408471", "2139015071" ], "abstract": [ "Travel providers such as airlines and on-line travel agents are becoming more and more interested in understanding how passengers choose among alternative itineraries when searching for flights. This knowledge helps them better display and adapt their offer, taking into account market conditions and customer needs. Some common applications are not only filtering and sorting alternatives, but also changing certain attributes in real-time (e.g., changing the price). In this paper, we concentrate with the problem of modeling air passenger choices of flight itineraries. This problem has historically been tackled using classical Discrete Choice Modelling techniques. Traditional statistical approaches, in particular the Multinomial Logit model (MNL), is widely used in industrial applications due to its simplicity and general good performance. However, MNL models present several shortcomings and assumptions that might not hold in real applications. To overcome these difficulties, we present a new choice model based on Pointer Networks. Given an input sequence, this type of deep neural architecture combines Recurrent Neural Networks with the Attention Mechanism to learn the conditional probability of an output whose values correspond to positions in an input sequence. Therefore, given a sequence of different alternatives presented to a customer, the model can learn to point to the one most likely to be chosen by the customer. The proposed method was evaluated on a real dataset that combines on-line user search logs and airline flight bookings. Experimental results show that the proposed model outperforms the traditional MNL model on several metrics.", "", "At the heart of Air Traffic Management (ATM) lies the Decision Support Systems (DST) that rely upon accurate trajectory prediction to determine how the airspace will look like in the future to make better decisions and advisories. Dealing with airspace that is prone to congestion due to environmental factors still remains the challenge especially when a deterministic approach is used in the trajectory prediction process. In this paper, we describe a novel stochastic trajectory prediction approach for ATM that can be used for more efficient and realistic flight planning and to assist airspace flow management, potentially resulting in higher safety, capacity, and efficiency commensurate with fuel savings thereby reducing emissions for a better environment. Our approach considers airspace as a 3D grid network, where each grid point is a location of a weather observation. We hypothetically build cubes around these grid points, so the entire airspace can be considered as a set of cubes. Each cube is defined by its centroid, the original grid point, and associated weather parameters that remain homogeneous within the cube during a period of time. Then, we align raw trajectories to a set of cube centroids which are basically fixed 3D positions independent of trajectory data. This creates a new form of trajectories which are 4D joint cubes, where each cube is a segment that is associated with not only spatio-temporal attributes but also with weather parameters. Next, we exploit machine learning techniques to train inference models from historical data and apply a stochastic model, a Hidden Markov Model (HMM), to predict trajectories taking environmental uncertainties into account. During the process, we apply time series clustering to generate input observations from an excessive set of weather parameters to feed into the Viterbi algorithm. Our experiments use a real trajectory dataset with pertaining weather observations and demonstrate the effectiveness of our approach to the trajectory prediction process for ATM.", "The systematic study of large-scale networks has unveiled the ubiquitous presence of connectivity patterns characterized by large-scale heterogeneities and unbounded statistical fluctuations. These features affect dramatically the behavior of the diffusion processes occurring on networks, determining the ensuing statistical properties of their evolution pattern and dynamics. In this article, we present a stochastic computational framework for the forecast of global epidemics that considers the complete worldwide air travel infrastructure complemented with census population data. We address two basic issues in global epidemic modeling: (i) we study the role of the large scale properties of the airline transportation network in determining the global diffusion pattern of emerging diseases; and (ii) we evaluate the reliability of forecasts and outbreak scenarios with respect to the intrinsic stochasticity of disease transmission and traffic flows. To address these issues we define a set of quantitative measures able to characterize the level of heterogeneity and predictability of the epidemic pattern. These measures may be used for the analysis of containment policies and epidemic risk assessment.", "A Revolution in the Making Method or Madness? The Computer Did It How It All Works When Passengers Collide Hold Me, Darlin' Upon Arrival: Hotels, Rental Cars, Cruise Ships, and More The Just Price The Scientists The State of Pricing Pricing's Many Faces The Coming Revolution", "As product prices become increasingly available on the World Wide Web, consumers attempt to understand how corporations vary these prices over time. However, corporations change prices based on proprietary algorithms and hidden variables (e.g., the number of unsold seats on a flight). Is it possible to develop data mining techniques that will enable consumers to predict price changes under these conditions?This paper reports on a pilot study in the domain of airline ticket prices where we recorded over 12,000 price observations over a 41 day period. When trained on this data, Hamlet --- our multi-strategy data mining algorithm --- generated a predictive model that saved 341 simulated passengers @math 320,572 in our simulation, thus H AMLET 's savings were 61.8 of optimal. The algorithm's savings of $198,074 represents an average savings of 23.8 for the 341 passengers for whom savings are possible. Overall, HAMLET saved 4.4 of the ticket price averaged over the entire set of 4,488 simulated passengers. Our pilot study suggests that mining of price data available over the web has the potential to save consumers substantial sums of money per annum." ] }
1812.01735	2950748128	We consider the problem of efficiently constructing cheap and novel round trip flight itineraries by combining legs from different airlines. We analyse the factors that contribute towards the price of such itineraries and find that many result from the combination of just 30 of airlines and that the closer the departure of such itineraries is to the user's search date the more likely they are to be cheaper than the tickets from one airline. We use these insights to formulate the problem as a trade-off between the recall of cheap itinerary constructions and the costs associated with building them. We propose a supervised learning solution with location embeddings which achieves an AUC=80.48, a substantial improvement over simpler baselines. We discuss various practical considerations for dealing with the staleness and the stability of the model and present the design of the machine learning pipeline. Finally, we present an analysis of the model's performance in production and its impact on Skyscanner's users.	Traditional ways to model airline prices have been based on complex networks @cite_4 @cite_15 or various supervised machine learning models @cite_10 @cite_18 . A more recent trend is around incorporating neural embeddings to model location data. Embeddings have seen great success in natural language processing @cite_9 , modelling large graphs @cite_0 and there has been a spike of enthusiasm around applying neural embedding to geographic location context with a variety of papers focusing on: a) mining embeddings from sequences of locations @cite_13 @cite_8 @cite_17 @cite_8 ; b) modelling geographic context @cite_20 @cite_7 @cite_6 and c) using alternative neural architectures where location representations are learned while optimising towards particular applications @cite_20 and different approaches are mixed together in @cite_6 and @cite_7 . The practicalities of augmenting existing non-deep machine learning pipelines with neural embeddings are discussed in @cite_11 and in @cite_12 .	{ "cite_N": [ "@cite_13", "@cite_18", "@cite_11", "@cite_4", "@cite_7", "@cite_8", "@cite_9", "@cite_6", "@cite_0", "@cite_15", "@cite_10", "@cite_12", "@cite_20", "@cite_17" ], "mid": [ "2951904259", "2743586218", "2605225344", "", "", "2964057288", "", "2763768291", "2154851992", "2043890595", "2139015071", "2595177306", "2776890924", "2573719245" ], "abstract": [ "Online social networks being extended to geographical space has resulted in large amount of user check-in data. Understanding check-ins can help to build appealing applications, such as location recommendation. In this paper, we propose DeepCity, a feature learning framework based on deep learning, to profile users and locations, with respect to user demographic and location category prediction. Both of the predictions are essential for social network companies to increase user engagement. The key contribution of DeepCity is the proposal of task-specific random walk which uses the location and user properties to guide the feature learning to be specific to each prediction task. Experiments conducted on 42M check-ins in three cities collected from Instagram have shown that DeepCity achieves a superior performance and outperforms other baseline models significantly.", "Travel providers such as airlines and on-line travel agents are becoming more and more interested in understanding how passengers choose among alternative itineraries when searching for flights. This knowledge helps them better display and adapt their offer, taking into account market conditions and customer needs. Some common applications are not only filtering and sorting alternatives, but also changing certain attributes in real-time (e.g., changing the price). In this paper, we concentrate with the problem of modeling air passenger choices of flight itineraries. This problem has historically been tackled using classical Discrete Choice Modelling techniques. Traditional statistical approaches, in particular the Multinomial Logit model (MNL), is widely used in industrial applications due to its simplicity and general good performance. However, MNL models present several shortcomings and assumptions that might not hold in real applications. To overcome these difficulties, we present a new choice model based on Pointer Networks. Given an input sequence, this type of deep neural architecture combines Recurrent Neural Networks with the Attention Mechanism to learn the conditional probability of an output whose values correspond to positions in an input sequence. Therefore, given a sequence of different alternatives presented to a customer, the model can learn to point to the one most likely to be chosen by the customer. The proposed method was evaluated on a real dataset that combines on-line user search logs and airline flight bookings. Experimental results show that the proposed model outperforms the traditional MNL model on several metrics.", "Deep Neural Networks (DNN) have demonstrated superior ability to extract high level embedding vectors from low level features. Despite the success, the serving time is still the bottleneck due to expensive run-time computation of multiple layers of dense matrices. GPGPU, FPGA, or ASIC-based serving systems require additional hardware that are not in the mainstream design of most commercial applications. In contrast, tree or forest-based models are widely adopted because of low serving cost, but heavily depend on carefully engineered features. This work proposes a Deep Embedding Forest model that benefits from the best of both worlds. The model consists of a number of embedding layers and a forest tree layer. The former maps high dimensional (hundreds of thousands to millions) and heterogeneous low-level features to the lower dimensional (thousands) vectors, and the latter ensures fast serving. Built on top of a representative DNN model called Deep Crossing, and two forest tree-based models including XGBoost and LightGBM, a two-step Deep Embedding Forest algorithm is demonstrated to achieve on-par or slightly better performance as compared with the DNN counterpart, with only a fraction of serving time on conventional hardware. After comparing with a joint optimization algorithm called partial fuzzification, also proposed in this paper, it is concluded that the two-step Deep Embedding Forest has achieved near optimal performance. Experiments based on large scale data sets (up to 1 billion samples) from a major sponsored search engine proves the efficacy of the proposed model.", "", "", "Point-of-interest (POI) recommendation is an important application for location-based social networks (LBSNs), which learns the user preference and mobility pattern from check-in sequences to recommend POIs. Previous studies show that modeling the sequential pattern of user check-ins is necessary for POI recommendation. Markov chain model, recurrent neural network, and the word2vec framework are used to model check-in sequences in previous work. However, all previous sequential models ignore the fact that check-in sequences on different days naturally exhibit the various temporal characteristics, for instance, \"work\" on weekday and \"entertainment\" on weekend. In this paper, we take this challenge and propose a Geo-Temporal sequential embedding rank (Geo-Teaser) model for POI recommendation. Inspired by the success of the word2vec framework to model the sequential contexts, we propose a temporal POI embedding model to learn POI representations under some particular temporal state. The temporal POI embedding model captures the contextual check-in information in sequences and the various temporal characteristics on different days as well. Furthermore, We propose a new way to incorporate the geographical influence into the pairwise preference ranking method through discriminating the unvisited POIs according to geographical information. Then we develop a geographically hierarchical pairwise preference ranking model. Finally, we propose a unified framework to recommend POIs combining these two models. To verify the effectiveness of our proposed method, we conduct experiments on two real-life datasets. Experimental results show that the Geo-Teaser model outperforms state-of-the-art models. Compared with the best baseline competitor, the Geo-Teaser model improves at least 20 on both datasets for all metrics.", "", "The application of neural embedding algorithms (based on architectures like skip-grams) to large knowledge bases like Wikipedia and the Google News Corpus has tremendously benefited multiple communities in applications as diverse as sentiment analysis, named entity recognition and text classification. In this paper, we present a similar resource for geospatial applications. We systematically construct a weighted network that spans all populated places in Geonames. Using a network embedding algorithm that was recently found to achieve excellent results and is based on the skip-gram model, we embed each populated place into a 100-dimensional vector space, in a similar vein as the GloVe embeddings released for Wikipedia. We demonstrate potential applications of this dataset resource, which we release under a public license.", "We present DeepWalk, a novel approach for learning latent representations of vertices in a network. These latent representations encode social relations in a continuous vector space, which is easily exploited by statistical models. DeepWalk generalizes recent advancements in language modeling and unsupervised feature learning (or deep learning) from sequences of words to graphs. DeepWalk uses local information obtained from truncated random walks to learn latent representations by treating walks as the equivalent of sentences. We demonstrate DeepWalk's latent representations on several multi-label network classification tasks for social networks such as BlogCatalog, Flickr, and YouTube. Our results show that DeepWalk outperforms challenging baselines which are allowed a global view of the network, especially in the presence of missing information. DeepWalk's representations can provide F1 scores up to 10 higher than competing methods when labeled data is sparse. In some experiments, DeepWalk's representations are able to outperform all baseline methods while using 60 less training data. DeepWalk is also scalable. It is an online learning algorithm which builds useful incremental results, and is trivially parallelizable. These qualities make it suitable for a broad class of real world applications such as network classification, and anomaly detection.", "The systematic study of large-scale networks has unveiled the ubiquitous presence of connectivity patterns characterized by large-scale heterogeneities and unbounded statistical fluctuations. These features affect dramatically the behavior of the diffusion processes occurring on networks, determining the ensuing statistical properties of their evolution pattern and dynamics. In this article, we present a stochastic computational framework for the forecast of global epidemics that considers the complete worldwide air travel infrastructure complemented with census population data. We address two basic issues in global epidemic modeling: (i) we study the role of the large scale properties of the airline transportation network in determining the global diffusion pattern of emerging diseases; and (ii) we evaluate the reliability of forecasts and outbreak scenarios with respect to the intrinsic stochasticity of disease transmission and traffic flows. To address these issues we define a set of quantitative measures able to characterize the level of heterogeneity and predictability of the epidemic pattern. These measures may be used for the analysis of containment policies and epidemic risk assessment.", "As product prices become increasingly available on the World Wide Web, consumers attempt to understand how corporations vary these prices over time. However, corporations change prices based on proprietary algorithms and hidden variables (e.g., the number of unsold seats on a flight). Is it possible to develop data mining techniques that will enable consumers to predict price changes under these conditions?This paper reports on a pilot study in the domain of airline ticket prices where we recorded over 12,000 price observations over a 41 day period. When trained on this data, Hamlet --- our multi-strategy data mining algorithm --- generated a predictive model that saved 341 simulated passengers @math 320,572 in our simulation, thus H AMLET 's savings were 61.8 of optimal. The algorithm's savings of $198,074 represents an average savings of 23.8 for the 341 passengers for whom savings are possible. Overall, HAMLET saved 4.4 of the ticket price averaged over the entire set of 4,488 simulated passengers. Our pilot study suggests that mining of price data available over the web has the potential to save consumers substantial sums of money per annum.", "We describe the Customer LifeTime Value (CLTV) prediction system deployed at ASOS.com, a global online fashion retailer. CLTV prediction is an important problem in e-commerce where an accurate estimate of future value allows retailers to effectively allocate marketing spend, identify and nurture high value customers and mitigate exposure to losses. The system at ASOS provides daily estimates of the future value of every customer and is one of the cornerstones of the personalised shopping experience. The state of the art in this domain uses large numbers of handcrafted features and ensemble regressors to forecast value, predict churn and evaluate customer loyalty. Recently, domains including language, vision and speech have shown dramatic advances by replacing handcrafted features with features that are learned automatically from data. We detail the system deployed at ASOS and show that learning feature representations is a promising extension to the state of the art in CLTV modelling. We propose a novel way to generate embeddings of customers, which addresses the issue of the ever changing product catalogue and obtain a significant improvement over an exhaustive set of handcrafted features.", "Understanding, representing, and reasoning about Points Of Interest (POI) types such as Auto Repair, Body Shop, Gas Stations, or Planetarium, is a key aspect of geographic information retrieval, recommender systems, geographic knowledge graphs, as well as studying urban spaces in general, e.g., for extracting functional or vague cognitive regions from user-generated content. One prerequisite to these tasks is the ability to capture the similarity and relatedness between POI types. Intuitively, a spatial search that returns body shops or even gas stations in the absence of auto repair places is still likely to satisfy some user needs while returning planetariums will not. Place hierarchies are frequently used for query expansion, but most of the existing hierarchies are relatively shallow and structured from a single perspective, thereby putting POI types that may be closely related regarding some characteristics far apart from another. This leads to the question of how to learn POI type representations from data. Models such as Word2Vec that produces word embeddings from linguistic contexts are a novel and promising approach as they come with an intuitive notion of similarity. However, the structure of geographic space, e.g., the interactions between POI types, differs substantially from linguistics. In this work, we present a novel method to augment the spatial contexts of POI types using a distance-binned, information-theoretic approach to generate embeddings. We demonstrate that our work outperforms Word2Vec and other models using three different evaluation tasks and strongly correlates with human assessments of POI type similarity. We published the resulting embeddings for 570 place types as well as a collection of human similarity assessments online for others to use.", "Conventional location recommendation models rely on users' visit history, geographical influence, temporal influence, etc., to infer users' preferences for locations. However, systematically modeling a location's context (i.e., the set of locations visited before or after this location) is relatively unexplored. In this paper, by leveraging the Skipgram model, we learn the latent representation for a location to capture the influence of its context. A pair-wise ranking loss that considers the confidences of observed user preferences for locations is then proposed to learn users' latent representations for personalized top-N location recommendations. Moreover, we also extend our model by taking into account temporal influence. Stochastic gradient descent based optimization algorithms are developed to fit the models. We conduct comprehensive experiments over four real datasets. Experimental results demonstrate that our approach significantly outperforms the state-of-the-art location recommendation methods." ] }
1812.01735	2950748128	We consider the problem of efficiently constructing cheap and novel round trip flight itineraries by combining legs from different airlines. We analyse the factors that contribute towards the price of such itineraries and find that many result from the combination of just 30 of airlines and that the closer the departure of such itineraries is to the user's search date the more likely they are to be cheaper than the tickets from one airline. We use these insights to formulate the problem as a trade-off between the recall of cheap itinerary constructions and the costs associated with building them. We propose a supervised learning solution with location embeddings which achieves an AUC=80.48, a substantial improvement over simpler baselines. We discuss various practical considerations for dealing with the staleness and the stability of the model and present the design of the machine learning pipeline. Finally, we present an analysis of the model's performance in production and its impact on Skyscanner's users.	The research community has recently started recognising the importance of sharing experience and learning in the way machine learning and data mining systems are implemented in production systems. In @cite_14 the authors stress the importance of investing considerable thinking and resources in building long-lasting technological infrastructures for machine learning systems. The authors of @cite_3 describe their experiences in building a recommendation engine, providing a great summary of business and technological constraints in which machine learning researchers and engineers operate when working on production systems. In @cite_2 the developers of Google Drive share their experience on the importance of reconsidering UI metrics and launch strategies for online experimentation with new machine learning features. Alibaba research in @cite_1 emphasises the importance of considering performance constraints and user experience and feedback in addition to accuracy when deploying machine learning in production.	{ "cite_N": [ "@cite_14", "@cite_1", "@cite_3", "@cite_2" ], "mid": [ "2189162242", "2622338386", "2593507512", "2714259754" ], "abstract": [ "Machine learning offers a fantastically powerful toolkit for building useful complex prediction systems quickly. This paper argues it is dangerous to think of these quick wins as coming for free. Using the software engineering framework of technical debt, we find it is common to incur massive ongoing maintenance costs in real-world ML systems. We explore several ML-specific risk factors to account for in system design. These include boundary erosion, entanglement, hidden feedback loops, undeclared consumers, data dependencies, configuration issues, changes in the external world, and a variety of system-level anti-patterns.", "In the 'Big Data' era, many real-world applications like search involve the ranking problem for a large number of items. It is important to obtain effective ranking results and at the same time obtain the results efficiently in a timely manner for providing good user experience and saving computational costs. Valuable prior research has been conducted for learning to efficiently rank like the cascade ranking (learning) model, which uses a sequence of ranking functions to progressively filter some items and rank the remaining items. However, most existing research of learning to efficiently rank in search is studied in a relatively small computing environments with simulated user queries. This paper presents novel research and thorough study of designing and deploying a Cascade model in a Large-scale Operational E-commerce Search application (CLOES), which deals with hundreds of millions of user queries per day with hundreds of servers. The challenge of the real-world application provides new insights for research: 1). Real-world search applications often involve multiple factors of preferences or constraints with respect to user experience and computational costs such as search accuracy, search latency, size of search results and total CPU cost, while most existing search solutions only address one or two factors; 2). Effectiveness of e-commerce search involves multiple types of user behaviors such as click and purchase, while most existing cascade ranking in search only models the click behavior. Based on these observations, a novel cascade ranking model is designed and deployed in an operational e-commerce search application. An extensive set of experiments demonstrate the advantage of the proposed work to address multiple factors of effectiveness, efficiency and user experience in the real-world application.", "Related Pins is the Web-scale recommender system that powers over 40 of user engagement on Pinterest. This paper is a longitudinal study of three years of its development, exploring the evolution of the system and its components from prototypes to present state. Each component was originally built with many constraints on engineering effort and computational resources, so we prioritized the simplest and highest-leverage solutions. We show how organic growth led to a complex system and how we managed this complexity. Many challenges arose while building this system, such as avoiding feedback loops, evaluating performance, activating content, and eliminating legacy heuristics. Finally, we offer suggestions for tackling these challenges when engineering Web-scale recommender systems.", "Google Drive is a cloud storage and collaboration service used by hundreds of millions of users around the world. Quick Access is a new feature in Google Drive that surfaces the most relevant documents when a user visits the home screen. Our metrics show that users locate their documents in half the time with this feature compared to previous approaches. The development of Quick Access illustrates many general challenges and constraints associated with practical machine learning such as protecting user privacy, working with data services that are not designed with machine learning in mind, and evolving product definitions. We believe that the lessons learned from this experience will be useful to practitioners tackling a wide range of applied machine learning problems." ] }
1812.01752	2959642114	Deep learning has been shown to produce state of the art results in many tasks in biomedical imaging, especially in segmentation. Moreover, segmentation of the cerebrovascular structure from magnetic resonance angiography is a challenging problem because its complex geometry and topology have a large inter-patient variability. Therefore, in this work, we present a convolutional neural network approach for this problem. Particularly, a new network topology inspired by the U-net 3D and by the Inception modules, entitled Uception. In addition, a discussion about the best objective function for sparse data also guided most choices during the project. State of the art models are also implemented for a comparison purpose and final results show that the proposed architecture has the best performance in this particular context.	In the field of 3D MRA imaging, several methods have been proposed to segment the vascular network: region-growing methods, differential analysis, model-based filtering, deformable models, path finding, mathematical morphology methods and hybrids between those methods @cite_9 . All these methods take into account the fact that vessels are thin and elongated structures. Recently, deep learning techniques have been used to segment 3D in vivo multiphoton images of vasculature @cite_0 or retinal blood vessels @cite_2 .	{ "cite_N": [ "@cite_0", "@cite_9", "@cite_2" ], "mid": [ "2781824996", "173412550", "2327793514" ], "abstract": [ "The health and function of tissue rely on its vasculature network to provide reliable blood perfusion. Volumetric imaging approaches, such as multiphoton microscopy, are able to generate detailed 3D images of blood vessels that could contribute to our understanding of the role of vascular structure in normal physiology and in disease mechanisms. The segmentation of vessels, a core image analysis problem, is a bottleneck that has prevented the systematic comparison of 3D vascular architecture across experimental populations. We explored the use of convolutional neural networks to segment 3D vessels within volumetric in vivo images acquired by multiphoton microscopy. We evaluated different network architectures and machine learning techniques in the context of this segmentation problem. We show that our optimized convolutional neural network architecture, which we call DeepVess, yielded a segmentation accuracy that was better than both the current state-of-the-art and a trained human annotator, while also being orders of magnitude faster. To explore the effects of aging and Alzheimer's disease on capillaries, we applied DeepVess to 3D images of cortical blood vessels in young and old mouse models of Alzheimer's disease and wild type littermates. We found little difference in the distribution of capillary diameter or tortuosity between these groups, but did note a decrease in the number of longer capillary segments ( @math ) in aged animals as compared to young, in both wild type and Alzheimer's disease mouse models.", "In the last 20 years, progress in 3D medical imaging (such as MRI and CT) has led to the development of modalities devoted to visualise vascular structures. These angiographic images progressively proved their usefulness in the context of various clinical applications. However, such data are generally complex to analyse due to their size and low amount of relevant (vascular) information versus noise, artifacts and other anatomical structures. Therefore, there is an ongoing necessity to provide tools facilitating image visualisation and analysis. In this chapter, we first focus on vascular image analysis. In particular, we present a survey on both standard and recent vessel segmentation methodologies. We then discuss the existing ways to model anatomical knowledge via the computation of vascular atlases. Such atlases can notably be embedded in computer-aided radiology tools.", "The condition of the vascular network of human eye is an important diagnostic factor in ophthalmology. Its segmentation in fundus imaging is a nontrivial task due to variable size of vessels, relatively low contrast, and potential presence of pathologies like microaneurysms and hemorrhages. Many algorithms, both unsupervised and supervised, have been proposed for this purpose in the past. We propose a supervised segmentation technique that uses a deep neural network trained on a large (up to 400 @math 000) sample of examples preprocessed with global contrast normalization, zero-phase whitening, and augmented using geometric transformations and gamma corrections. Several variants of the method are considered, including structured prediction, where a network classifies multiple pixels simultaneously. When applied to standard benchmarks of fundus imaging, the DRIVE, STARE, and CHASE databases, the networks significantly outperform the previous algorithms on the area under ROC curve measure (up to @math ) and accuracy of classification (up to @math ). The method is also resistant to the phenomenon of central vessel reflex, sensitive in detection of fine vessels ( @math ), and fares well on pathological cases." ] }
1812.01752	2959642114	Deep learning has been shown to produce state of the art results in many tasks in biomedical imaging, especially in segmentation. Moreover, segmentation of the cerebrovascular structure from magnetic resonance angiography is a challenging problem because its complex geometry and topology have a large inter-patient variability. Therefore, in this work, we present a convolutional neural network approach for this problem. Particularly, a new network topology inspired by the U-net 3D and by the Inception modules, entitled Uception. In addition, a discussion about the best objective function for sparse data also guided most choices during the project. State of the art models are also implemented for a comparison purpose and final results show that the proposed architecture has the best performance in this particular context.	However, the use of deep learning in the problem of vascular segmentation from 3D MRA has been only recently proposed in @cite_12 , where the authors used cross-hair filters to do the convolutions, demanding less memory and computing over bigger volumes. They also used some balanced cross-entropy functions and fine-tuning after an initial training with synthetic data.	{ "cite_N": [ "@cite_12" ], "mid": [ "2794642825" ], "abstract": [ "We present DeepVesselNet, an architecture tailored to the challenges faced when extracting vessel networks or trees and corresponding features in 3-D angiographic volumes using deep learning. We discuss the problems of low execution speed and high memory requirements associated with full 3-D convolutional networks, high-class imbalance arising from the low percentage of vessel voxels, and unavailability of accurately annotated training data - and offer solutions as the building blocks of DeepVesselNet. First, we formulate 2-D orthogonal cross-hair filters which make use of 3-D context information at a reduced computational burden. Second, we introduce a class balancing cross-entropy loss function with false positive rate correction to handle the high-class imbalance and high false positive rate problems associated with existing loss functions. Finally, we generate synthetic dataset using a computational angiogenesis model capable of generating vascular trees under physiological constraints on local network structure and topology and use these data for transfer learning. DeepVesselNet is optimized for segmenting and analyzing vessels, and we test the performance on a range of angiographic volumes including clinical MRA data of the human brain, as well as X-ray tomographic microscopy scans of the rat brain. Our experiments show that, by replacing 3-D filters with cross-hair filters in our network, we achieve over 23 improvement in speed, lower memory footprint, lower network complexity which prevents overfitting and comparable accuracy (with a Cox-Wilcoxon paired sample significance test p-value of 0.07 when compared to full 3-D filters). Our class balancing metric is crucial for training the network and transfer learning with synthetic data is an efficient, robust, and very generalizable approach leading to a network that excels in a variety of angiography segmentation tasks." ] }
1812.01752	2959642114	Deep learning has been shown to produce state of the art results in many tasks in biomedical imaging, especially in segmentation. Moreover, segmentation of the cerebrovascular structure from magnetic resonance angiography is a challenging problem because its complex geometry and topology have a large inter-patient variability. Therefore, in this work, we present a convolutional neural network approach for this problem. Particularly, a new network topology inspired by the U-net 3D and by the Inception modules, entitled Uception. In addition, a discussion about the best objective function for sparse data also guided most choices during the project. State of the art models are also implemented for a comparison purpose and final results show that the proposed architecture has the best performance in this particular context.	More generally, recent segmentation techniques based on deep learning use end-to-end CNNs, where there are no dense layers. One of the most notable is the U-net @cite_1 , where there is an encoding path with series of convolutional layers and max-pooling layers, and a decoder path, with a series of up-sampling and convolutional layers, and between them, there are shortcut connections which make sure that the spatial information is being passed to the output. The same model was also later designed with 3D convolutions @cite_10 .	{ "cite_N": [ "@cite_10", "@cite_1" ], "mid": [ "2464708700", "2952232639" ], "abstract": [ "This paper introduces a network for volumetric segmentation that learns from sparsely annotated volumetric images. We outline two attractive use cases of this method: (1) In a semi-automated setup, the user annotates some slices in the volume to be segmented. The network learns from these sparse annotations and provides a dense 3D segmentation. (2) In a fully-automated setup, we assume that a representative, sparsely annotated training set exists. Trained on this data set, the network densely segments new volumetric images. The proposed network extends the previous u-net architecture from by replacing all 2D operations with their 3D counterparts. The implementation performs on-the-fly elastic deformations for efficient data augmentation during training. It is trained end-to-end from scratch, i.e., no pre-trained network is required. We test the performance of the proposed method on a complex, highly variable 3D structure, the Xenopus kidney, and achieve good results for both use cases.", "There is large consent that successful training of deep networks requires many thousand annotated training samples. In this paper, we present a network and training strategy that relies on the strong use of data augmentation to use the available annotated samples more efficiently. The architecture consists of a contracting path to capture context and a symmetric expanding path that enables precise localization. We show that such a network can be trained end-to-end from very few images and outperforms the prior best method (a sliding-window convolutional network) on the ISBI challenge for segmentation of neuronal structures in electron microscopic stacks. Using the same network trained on transmitted light microscopy images (phase contrast and DIC) we won the ISBI cell tracking challenge 2015 in these categories by a large margin. Moreover, the network is fast. Segmentation of a 512x512 image takes less than a second on a recent GPU. The full implementation (based on Caffe) and the trained networks are available at this http URL ." ] }
1812.01752	2959642114	Deep learning has been shown to produce state of the art results in many tasks in biomedical imaging, especially in segmentation. Moreover, segmentation of the cerebrovascular structure from magnetic resonance angiography is a challenging problem because its complex geometry and topology have a large inter-patient variability. Therefore, in this work, we present a convolutional neural network approach for this problem. Particularly, a new network topology inspired by the U-net 3D and by the Inception modules, entitled Uception. In addition, a discussion about the best objective function for sparse data also guided most choices during the project. State of the art models are also implemented for a comparison purpose and final results show that the proposed architecture has the best performance in this particular context.	Other similar networks in 3D, like the V-net, uses convolutions with strides of 2 instead of max-pooling and residual connections to improve convergence time @cite_3 .	{ "cite_N": [ "@cite_3" ], "mid": [ "2432481613" ], "abstract": [ "Convolutional Neural Networks (CNNs) have been recently employed to solve problems from both the computer vision and medical image analysis fields. Despite their popularity, most approaches are only able to process 2D images while most medical data used in clinical practice consists of 3D volumes. In this work we propose an approach to 3D image segmentation based on a volumetric, fully convolutional, neural network. Our CNN is trained end-to-end on MRI volumes depicting prostate, and learns to predict segmentation for the whole volume at once. We introduce a novel objective function, that we optimise during training, based on Dice coefficient. In this way we can deal with situations where there is a strong imbalance between the number of foreground and background voxels. To cope with the limited number of annotated volumes available for training, we augment the data applying random non-linear transformations and histogram matching. We show in our experimental evaluation that our approach achieves good performances on challenging test data while requiring only a fraction of the processing time needed by other previous methods." ] }
1812.01923	2949253602	Detecting unintended falls is essential for ambient intelligence and healthcare of elderly people living alone. In recent years, deep convolutional nets are widely used in human action analysis, based on which a number of fall detection methods have been proposed. Despite their highly effective performances, the behaviors of how the convolutional nets recognize falls are still not clear. In this paper, instead of proposing a novel approach, we perform a systematical empirical study, attempting to investigate the underlying fall recognition process. We propose four tasks to investigate, which involve five types of input modalities, seven net instances and different training samples. The obtained quantitative and qualitative results reveal the patterns that the nets tend to learn, and several factors that can heavily influence the performances on fall recognition. We expect that our conclusions are favorable to proposing better deep learning solutions to fall detection systems.	Systematic reviews of fall recognition and detection systems can be found in @cite_43 and @cite_2 , which cover solutions based on diverse types of sensors. For vision-based methods, a typical processing pipeline consists of background subtraction, feature extraction and classification, as presented in @cite_23 @cite_41 and others. Each step in this pipeline is normally hand-crafted, separately considered and implemented based on certain heuristic rules. A frequently considered rule is that the background information is redundant for fall detection. Thus, background subtraction is performed by algorithms like training Gaussian mixture models, subspace clustering or other sophisticated approaches @cite_21 . Another heuristic rule is that the body shape is a pronounced feature of falling. Consequently, the silhouette of the human body @cite_23 @cite_24 , or the shape of the foreground bounding box @cite_41 @cite_28 , is extracted and analyzed. Nevertheless, heuristics are not always precise and comprehensive. The studies of @cite_34 and @cite_4 present effective fall detection solutions when considering the ground plane, indicating that the background information can be very useful.	{ "cite_N": [ "@cite_4", "@cite_41", "@cite_28", "@cite_21", "@cite_24", "@cite_43", "@cite_23", "@cite_2", "@cite_34" ], "mid": [ "2738996446", "", "1930442888", "", "2131889717", "2155326828", "2096300942", "2076068958", "2100716893" ], "abstract": [ "Falls are serious and costly for elderly people. The Centers for Disease Control and Prevention of the US reports that millions of older people, 65 and older, fall each year at least once. Serious injuries such as; hip fractures, broken bones or head injury, are caused by 20 of the falls. The time it takes to respond and treat a fallen person is crucial. With this paper we present a new , non-invasive system for fallen people detection. Our approach uses only stereo camera data for passively sensing the environment. The key novelty is a human fall detector which uses a CNN based human pose estimator in combination with stereo data to reconstruct the human pose in 3D and estimate the ground plane in 3D. Furthermore, our system consists of a reasoning module which formulates a number of measures to reason whether a person is fallen. We have tested our approach in different scenarios covering most activities elderly people might encounter living at home. Based on our extensive evaluations, our systems shows high accuracy and almost no miss-classification. To reproduce our results, the implementation is publicly available to the scientific community.", "", "Automatic detection of a falling person in video is an important problem with applications in security and safety areas including supportive home environments and CCTV surveillance systems. Human motion in video is modeled using Hidden Markov Models (HMM) in this paper. In addition, the audio track of the video is also used to distinguish a person simply sitting on a floor from a person stumbling and falling. Most video recording systems have the capability of recording audio as well and the impact sound of a falling person is also available as an additional clue. Audio channel data based decision is also reached using HMMs and fused with results of HMMs modeling the video data to reach a final decision.", "", "A major problem among the elderly involves falling. The recognition of falls from video first requires the segmentation of the individual from the background. To ensure privacy, segmentation should result in a silhouette that is a binary map indicating only the body position of the individual in an image. We have previously demonstrated a segmentation method based on color that can recognize the silhouette and detect and remove shadows. After the silhouettes are obtained, we extract features and train hidden Markov models to recognize future performances of these known activities. In this paper, we present preliminary results that demonstrate the usefulness of this approach for distinguishing between a few common activities, specifically with fall detection in mind. and dynamic environments. High-level knowledge was fused with low-level feature-based classification to handle a time-varying background, and a decision process based on a fuzzy logic inference system was used to detach the moving objects from the human silhouette. One central task in silhouette extraction is background modeling. Once a background model is established, those image regions with significantly different characteristics from the background are considered foreground objects. In (5), a Least Median of Squares method was used to construct a background model, and in (6), a differencing function was proposed to extract moving human silhouettes. (7) demonstrated a non-parametric kernel density estimation method to model and subtract the human from a background. The adaptive background updating handles small motion in the background scene, such as moving tree branches. A slightly different approach outlined in (8) is based on processing in the YUV color space, and is capable of detecting shadows and extracting", "Since falls are a major public health problem among older people, the number of systems aimed at detecting them has increased dramatically over recent years. This work presents an extensive literature review of fall detection systems, including comparisons among various kinds of studies. It aims to serve as a reference for both clinicians and biomedical engineers planning or conducting field investigations. Challenges, issues and trends in fall detection have been identified after the reviewing work. The number of studies using context-aware techniques is still increasing but there is a new trend towards the integration of fall detection into smartphones as well as the use of machine learning methods in the detection algorithm. We have also identified challenges regarding performance under real-life conditions, usability, and user acceptance as well as issues related to power consumption, real-time operations, sensing limitations, privacy and record of real-life falls.", "Faced with the growing population of seniors, developed countries need to establish new healthcare systems to ensure the safety of elderly people at home. Computer vision provides a promising solution to analyze personal behavior and detect certain unusual events such as falls. In this paper, a new method is proposed to detect falls by analyzing human shape deformation during a video sequence. A shape matching technique is used to track the person's silhouette along the video sequence. The shape deformation is then quantified from these silhouettes based on shape analysis methods. Finally, falls are detected from normal activities using a Gaussian mixture model. This paper has been conducted on a realistic data set of daily activities and simulated falls, and gives very good results (as low as 0 error with a multi-camera setup) compared with other common image processing methods.", "Fall detection is a major challenge in the public health care domain, especially for the elderly, and reliable surveillance is a necessity to mitigate the effects of falls. The technology and products related to fall detection have always been in high demand within the security and the health-care industries. An effective fall detection system is required to provide urgent support and to significantly reduce the medical care costs associated with falls. In this paper, we give a comprehensive survey of different systems for fall detection and their underlying algorithms. Fall detection approaches are divided into three main categories: wearable device based, ambience device based and vision based. These approaches are summarised and compared with each other and a conclusion is derived with some discussions on possible future work.", "A method for detecting falls in the homes of older adults using the Microsoft Kinect and a two-stage fall detection system is presented. The first stage of the detection system characterizes a person's vertical state in individual depth image frames, and then segments on ground events from the vertical state time series obtained by tracking the person over time. The second stage uses an ensemble of decision trees to compute a confidence that a fall preceded on a ground event. Evaluation was conducted in the actual homes of older adults, using a combined nine years of continuous data collected in 13 apartments. The dataset includes 454 falls, 445 falls performed by trained stunt actors and nine naturally occurring resident falls. The extensive data collection allows for characterization of system performance under real-world conditions to a degree that has not been shown in other studies. Cross validation results are included for standing, sitting, and lying down positions, near (within 4 m) versus far fall locations, and occluded versus not occluded fallers. The method is compared against five state-of-the-art fall detection algorithms and significantly better results are achieved." ] }
1812.01923	2949253602	Detecting unintended falls is essential for ambient intelligence and healthcare of elderly people living alone. In recent years, deep convolutional nets are widely used in human action analysis, based on which a number of fall detection methods have been proposed. Despite their highly effective performances, the behaviors of how the convolutional nets recognize falls are still not clear. In this paper, instead of proposing a novel approach, we perform a systematical empirical study, attempting to investigate the underlying fall recognition process. We propose four tasks to investigate, which involve five types of input modalities, seven net instances and different training samples. The obtained quantitative and qualitative results reveal the patterns that the nets tend to learn, and several factors that can heavily influence the performances on fall recognition. We expect that our conclusions are favorable to proposing better deep learning solutions to fall detection systems.	Comparing with traditional vision-based approaches, deep learning methods enable end-to-end inference with minimal pre-processing on the input data, and the deep nets can learn representative features from the data automatically. Therefore, the algorithm is not necessary to rely on non-guaranteed heuristics. Several studies report that deep learning methods lead to better performances in terms of action recognition @cite_40 @cite_9 , action detection @cite_22 @cite_35 @cite_36 , action parsing @cite_5 @cite_32 and other tasks of human behavior analysis. Their success encourages many studies of fall recognition based on deep neural networks. For example, the work of @cite_26 employs a convolutional net with a similar architecture to the VGG-16 net @cite_7 and uses optical flow as the input modality. The work of @cite_15 uses a PCANet to recognize falls from image sequences with the assistance of foreground detection.	{ "cite_N": [ "@cite_35", "@cite_26", "@cite_22", "@cite_7", "@cite_36", "@cite_9", "@cite_32", "@cite_40", "@cite_5", "@cite_15" ], "mid": [ "1923332106", "2772973030", "2179401333", "1686810756", "", "2619082050", "", "", "2550143307", "657467088" ], "abstract": [ "We address the problem of action detection in videos. Driven by the latest progress in object detection from 2D images, we build action models using rich feature hierarchies derived from shape and kinematic cues. We incorporate appearance and motion in two ways. First, starting from image region proposals we select those that are motion salient and thus are more likely to contain the action. This leads to a significant reduction in the number of regions being processed and allows for faster computations. Second, we extract spatio-temporal feature representations to build strong classifiers using Convolutional Neural Networks. We link our predictions to produce detections consistent in time, which we call action tubes. We show that our approach outperforms other techniques in the task of action detection.", "One of the biggest challenges in modern societies is the improvement of healthy aging and the support to older persons in their daily activities. In particular, given its social and economic impact, the automatic detection of falls has attracted considerable attention in the computer vision and pattern recognition communities. Although the approaches based on wearable sensors have provided high detection rates, some of the potential users are reluctant to wear them and thus their use is not yet normalized. As a consequence, alternative approaches such as vision-based methods have emerged. We firmly believe that the irruption of the Smart Environments and the Internet of Things paradigms, together with the increasing number of cameras in our daily environment, forms an optimal context for vision-based systems. Consequently, here we propose a vision-based solution using Convolutional Neural Networks to decide if a sequence of frames contains a person falling. To model the video motion and make the system scenario independent, we use optical flow images as input to the networks followed by a novel three-step training phase. Furthermore, our method is evaluated in three public datasets achieving the state-of-the-art results in all three of them.", "In this work we introduce a fully end-to-end approach for action detection in videos that learns to directly predict the temporal bounds of actions. Our intuition is that the process of detecting actions is naturally one of observation and refinement: observing moments in video, and refining hypotheses about when an action is occurring. Based on this insight, we formulate our model as a recurrent neural network-based agent that interacts with a video over time. The agent observes video frames and decides both where to look next and when to emit a prediction. Since backpropagation is not adequate in this non-differentiable setting, we use REINFORCE to learn the agent's decision policy. Our model achieves state-of-the-art results on the THUMOS'14 and ActivityNet datasets while observing only a fraction (2 or less) of the video frames.", "In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.", "", "The paucity of videos in current action classification datasets (UCF-101 and HMDB-51) has made it difficult to identify good video architectures, as most methods obtain similar performance on existing small-scale benchmarks. This paper re-evaluates state-of-the-art architectures in light of the new Kinetics Human Action Video dataset. Kinetics has two orders of magnitude more data, with 400 human action classes and over 400 clips per class, and is collected from realistic, challenging YouTube videos. We provide an analysis on how current architectures fare on the task of action classification on this dataset and how much performance improves on the smaller benchmark datasets after pre-training on Kinetics. We also introduce a new Two-Stream Inflated 3D ConvNet (I3D) that is based on 2D ConvNet inflation: filters and pooling kernels of very deep image classification ConvNets are expanded into 3D, making it possible to learn seamless spatio-temporal feature extractors from video while leveraging successful ImageNet architecture designs and even their parameters. We show that, after pre-training on Kinetics, I3D models considerably improve upon the state-of-the-art in action classification, reaching 80.9 on HMDB-51 and 98.0 on UCF-101.", "", "", "The ability to identify and temporally segment fine-grained human actions throughout a video is crucial for robotics, surveillance, education, and beyond. Typical approaches decouple this problem by first extracting local spatiotemporal features from video frames and then feeding them into a temporal classifier that captures high-level temporal patterns. We describe a class of temporal models, which we call Temporal Convolutional Networks (TCNs), that use a hierarchy of temporal convolutions to perform fine-grained action segmentation or detection. Our Encoder-Decoder TCN uses pooling and upsampling to efficiently capture long-range temporal patterns whereas our Dilated TCN uses dilated convolutions. We show that TCNs are capable of capturing action compositions, segment durations, and long-range dependencies, and are over a magnitude faster to train than competing LSTM-based Recurrent Neural Networks. We apply these models to three challenging fine-grained datasets and show large improvements over the state of the art.", "Fall incidents have been reported as the second most common cause of death, especially for elderly people. Human fall detection is necessary in smart home healthcare systems. Recently various fall detection approaches have been proposed., among which computer vision based approaches offer a promising and effective way. In this paper, we proposed a new framework for fall detection based on automatic feature learning methods. First, the extracted frames, including human from video sequences of different views, form the training set. Then, a PCANet model is trained by using all samples to predict the label of every frame. Because a fall behavior is contained in many continuous frames, the reliable fall detection should not only analyze one frame but also a video sequence. Based on the prediction result of the trained PCANet model for each frame, an action model is further obtained by SVM with the predicted labels of frames in video sequences. Experiments show that the proposed method achieved reliable results compared with other commonly used methods based on the multiple cameras fall dataset, and a better result is further achieved in our dataset which contains more training samples." ] }
1812.01923	2949253602	Detecting unintended falls is essential for ambient intelligence and healthcare of elderly people living alone. In recent years, deep convolutional nets are widely used in human action analysis, based on which a number of fall detection methods have been proposed. Despite their highly effective performances, the behaviors of how the convolutional nets recognize falls are still not clear. In this paper, instead of proposing a novel approach, we perform a systematical empirical study, attempting to investigate the underlying fall recognition process. We propose four tasks to investigate, which involve five types of input modalities, seven net instances and different training samples. The obtained quantitative and qualitative results reveal the patterns that the nets tend to learn, and several factors that can heavily influence the performances on fall recognition. We expect that our conclusions are favorable to proposing better deep learning solutions to fall detection systems.	To understand the behaviors of deep convolutional nets, several types of attribute maps have been proposed @cite_38 @cite_8 . For a specific input and a target class, the attribute map has the same spatial resolution with the input, and reveals the influence of each input pixel to the probability of the target class. The work of @cite_11 proposes a saliency map, which is computed as the derivative of the output with respect to the input. @cite_18 proposes the integrated gradients , in which the values show the difference between the net output of a reference input (normally zero) and the net output of a sample. @cite_12 proposes the DeepLIFT attribute measure, which can be regarded as an approximated version of integrated gradients according to @cite_8 .	{ "cite_N": [ "@cite_38", "@cite_18", "@cite_8", "@cite_12", "@cite_11" ], "mid": [ "2769924204", "2949197630", "2953295770", "2952688545", "2962851944" ], "abstract": [ "The practical impact of deep learning on complex supervised learning problems has been significant, so much so that almost every Artificial Intelligence problem, or at least a portion thereof, has been somehow recast as a deep learning problem. The applications appeal is significant, but this appeal is increasingly challenged by what some call the challenge of explainability, or more generally the more traditional challenge of debuggability: if the outcomes of a deep learning process produce unexpected results (e.g., less than expected performance of a classifier), then there is little available in the way of theories or tools to help investigate the potential causes of such unexpected behavior, especially when this behavior could impact people's lives. We describe a preliminary framework to help address this issue, which we call \"deep visual explanation\" (DVE). \"Deep,\" because it is the development and performance of deep neural network models that we want to understand. \"Visual,\" because we believe that the most rapid insight into a complex multi-dimensional model is provided by appropriate visualization techniques, and \"Explanation,\" because in the spectrum from instrumentation by inserting print statements to the abductive inference of explanatory hypotheses, we believe that the key to understanding deep learning relies on the identification and exposure of hypotheses about the performance behavior of a learned deep model. In the exposition of our preliminary framework, we use relatively straightforward image classification examples and a variety of choices on initial configuration of a deep model building scenario. By careful but not complicated instrumentation, we expose classification outcomes of deep models using visualization, and also show initial results for one potential application of interpretability.", "We study the problem of attributing the prediction of a deep network to its input features, a problem previously studied by several other works. We identify two fundamental axioms---Sensitivity and Implementation Invariance that attribution methods ought to satisfy. We show that they are not satisfied by most known attribution methods, which we consider to be a fundamental weakness of those methods. We use the axioms to guide the design of a new attribution method called Integrated Gradients. Our method requires no modification to the original network and is extremely simple to implement; it just needs a few calls to the standard gradient operator. We apply this method to a couple of image models, a couple of text models and a chemistry model, demonstrating its ability to debug networks, to extract rules from a network, and to enable users to engage with models better.", "Understanding the flow of information in Deep Neural Networks (DNNs) is a challenging problem that has gain increasing attention over the last few years. While several methods have been proposed to explain network predictions, there have been only a few attempts to compare them from a theoretical perspective. What is more, no exhaustive empirical comparison has been performed in the past. In this work, we analyze four gradient-based attribution methods and formally prove conditions of equivalence and approximation between them. By reformulating two of these methods, we construct a unified framework which enables a direct comparison, as well as an easier implementation. Finally, we propose a novel evaluation metric, called Sensitivity-n and test the gradient-based attribution methods alongside with a simple perturbation-based attribution method on several datasets in the domains of image and text classification, using various network architectures.", "The purported \"black box\"' nature of neural networks is a barrier to adoption in applications where interpretability is essential. Here we present DeepLIFT (Deep Learning Important FeaTures), a method for decomposing the output prediction of a neural network on a specific input by backpropagating the contributions of all neurons in the network to every feature of the input. DeepLIFT compares the activation of each neuron to its 'reference activation' and assigns contribution scores according to the difference. By optionally giving separate consideration to positive and negative contributions, DeepLIFT can also reveal dependencies which are missed by other approaches. Scores can be computed efficiently in a single backward pass. We apply DeepLIFT to models trained on MNIST and simulated genomic data, and show significant advantages over gradient-based methods. A detailed video tutorial on the method is at this http URL and code is at this http URL.", "This paper addresses the visualisation of image classification models, learnt using deep Convolutional Networks (ConvNets). We consider two visualisation techniques, based on computing the gradient of the class score with respect to the input image. The first one generates an image, which maximises the class score [5], thus visualising the notion of the class, captured by a ConvNet. The second technique computes a class saliency map, specific to a given image and class. We show that such maps can be employed for weakly supervised object segmentation using classification ConvNets. Finally, we establish the connection between the gradient-based ConvNet visualisation methods and deconvolutional networks [13]." ] }
1812.01628	2950521938	Text-based adventure games provide a platform on which to explore reinforcement learning in the context of a combinatorial action space, such as natural language. We present a deep reinforcement learning architecture that represents the game state as a knowledge graph which is learned during exploration. This graph is used to prune the action space, enabling more efficient exploration. The question of which action to take can be reduced to a question-answering task, a form of transfer learning that pre-trains certain parts of our architecture. In experiments using the TextWorld framework, we show that our proposed technique can learn a control policy faster than baseline alternatives. We have also open-sourced our code at this https URL.	A growing body of research has explored the challenges associated with text-based games . @cite_3 attempt to solve parser-based text games by encoding the observations using an LSTM. This encoding vector is then used by an action scoring network that determines the scores for the action verb and each of the corresponding argument objects. The two scores are then averaged to determine @math -value for the state-action pair. @cite_0 present the Deep Reinforcement Relevance Network (DRRN) which uses two separate deep neural networks to encode the state and actions. The @math -value for a state-action pair is then computed by a pairwise interaction function between the two encoded representations. Both of these methods are not conditioned on previous observations and so are at a disadvantage when dealing with complex partially observable games. Additionally, neither of these approaches prune the action space and so end up wasting trials exploring state-action pairs that are likely to have low @math -values, likely leading to slower convergence times for combinatorially large action spaces.	{ "cite_N": [ "@cite_0", "@cite_3" ], "mid": [ "2964179661", "2112177991" ], "abstract": [ "This paper introduces a novel architecture for reinforcement learning with deep neural networks designed to handle state and action spaces characterized by natural language, as found in text-based games. Termed a deep reinforcement relevance network (DRRN), the architecture represents action and state spaces with separate embedding vectors, which are combined with an interaction function to approximate the Q-function in reinforcement learning. We evaluate the DRRN on two popular text games, showing superior performance over other deep Qlearning architectures. Experiments with paraphrased action descriptions show that the model is extracting meaning rather than simply memorizing strings of text.", "We present a general framework and learning algorithm for the task of concept labeling: each word in a given sentence has to be tagged with the unique physical entity (e.g. person, object or location) or abstract concept it refers to. Our method allows both world knowledge and linguistic information to be used during learning and prediction. We show experimentally that we can learn to use world knowledge to resolve ambiguities in language, such as word senses or reference resolution, without the use of handcrafted rules or features." ] }
1812.01628	2950521938	Text-based adventure games provide a platform on which to explore reinforcement learning in the context of a combinatorial action space, such as natural language. We present a deep reinforcement learning architecture that represents the game state as a knowledge graph which is learned during exploration. This graph is used to prune the action space, enabling more efficient exploration. The question of which action to take can be reduced to a question-answering task, a form of transfer learning that pre-trains certain parts of our architecture. In experiments using the TextWorld framework, we show that our proposed technique can learn a control policy faster than baseline alternatives. We have also open-sourced our code at this https URL.	Knowledge graphs have been demonstrated to improve natural language understanding in other domains outside of text adventure games. For example, @cite_2 use commonsense knowledge graphs such as ConceptNet to significantly improve the ability of neural networks to predict the end of a story. They represent the graph in terms of a knowledge context vector using features from ConceptNet and graph attention . The state representation that we have chosen as well as our method of action pruning builds on the strengths of existing approaches while simultaneously avoiding the shortcomings of ineffective exploration and lack of long-term context.	{ "cite_N": [ "@cite_2" ], "mid": [ "2952243835" ], "abstract": [ "Generating a reasonable ending for a given story context, i.e., story ending generation, is a strong indication of story comprehension. This task requires not only to understand the context clues which play an important role in planning the plot but also to handle implicit knowledge to make a reasonable, coherent story. In this paper, we devise a novel model for story ending generation. The model adopts an incremental encoding scheme to represent context clues which are spanning in the story context. In addition, commonsense knowledge is applied through multi-source attention to facilitate story comprehension, and thus to help generate coherent and reasonable endings. Through building context clues and using implicit knowledge, the model is able to produce reasonable story endings. context clues implied in the post and make the inference based on it. Automatic and manual evaluation shows that our model can generate more reasonable story endings than state-of-the-art baselines." ] }
1812.01742	2969824030	Single-view 3D shape reconstruction is an important but challenging problem, mainly for two reasons. First, as shape annotation is very expensive to acquire, current methods rely on synthetic data, in which ground-truth 3D annotation is easy to obtain. However, this results in domain adaptation problem when applied to natural images. The second challenge is that there are multiple shapes that can explain a given 2D image. In this paper, we propose a framework to improve over these challenges using adversarial training. On one hand, we impose domain confusion between natural and synthetic image representations to reduce the distribution gap. On the other hand, we impose the reconstruction to be realistic' by forcing it to lie on a (learned) manifold of realistic object shapes. Our experiments show that these constraints improve performance by a large margin over baseline reconstruction models. We achieve results competitive with the state of the art with a much simpler architecture.	Traditional reconstruction methods rely on multiple images of same object instance to achieve reconstruction @cite_50 @cite_5 @cite_49 @cite_21 @cite_38 . Recently, data-driven approaches to 3D reconstruction from single image have appeared. These methods can roughly be divided into two types: (i) those that explicitly use 3D structures @cite_36 @cite_44 @cite_37 @cite_3 @cite_28 @cite_15 and (ii) those that use other sources of information to infer the 3D structure @cite_20 @cite_4 @cite_35 @cite_39 @cite_26 @cite_49 @cite_48 @cite_29 .	{ "cite_N": [ "@cite_38", "@cite_37", "@cite_35", "@cite_4", "@cite_26", "@cite_36", "@cite_28", "@cite_48", "@cite_21", "@cite_29", "@cite_3", "@cite_39", "@cite_44", "@cite_49", "@cite_50", "@cite_5", "@cite_15", "@cite_20" ], "mid": [ "2257063750", "2546066744", "2950701417", "1893912098", "2619556892", "2964137676", "2963433432", "2963739349", "2160014001", "2895040926", "2560722161", "2963730200", "2342277278", "2116546169", "2117007522", "2030496844", "1920022804", "1977792424" ], "abstract": [ "This tutorial presents a hands-on view of the field of multi-view stereo with a focus on practical algorithms. Multi-view stereo algorithms are able to construct highly detailed 3D models from images alone. They take a possibly very large set of images and construct a 3D plausible geometry that explains the images under some reasonable assumptions, the most important being scene rigidity. The tutorial frames the multiview stereo problem as an image geometry consistency optimization problem. It describes in detail its main two ingredients: robust implementations of photometric consistency measures, and efficient optimization algorithms. It then presents how these main ingredients are used by some of the most successful algorithms, applied into real applications, and deployed as products in the industry. Finally it describes more advanced approaches exploiting domain-specific knowledge such as structural priors, and gives an overview of the remaining challenges and future research directions.", "We study the problem of 3D object generation. We propose a novel framework, namely 3D Generative Adversarial Network (3D-GAN), which generates 3D objects from a probabilistic space by leveraging recent advances in volumetric convolutional networks and generative adversarial nets. The benefits of our model are three-fold: first, the use of an adversarial criterion, instead of traditional heuristic criteria, enables the generator to capture object structure implicitly and to synthesize high-quality 3D objects; second, the generator establishes a mapping from a low-dimensional probabilistic space to the space of 3D objects, so that we can sample objects without a reference image or CAD models, and explore the 3D object manifold; third, the adversarial discriminator provides a powerful 3D shape descriptor which, learned without supervision, has wide applications in 3D object recognition. Experiments demonstrate that our method generates high-quality 3D objects, and our unsupervisedly learned features achieve impressive performance on 3D object recognition, comparable with those of supervised learning methods.", "Understanding the 3D world is a fundamental problem in computer vision. However, learning a good representation of 3D objects is still an open problem due to the high dimensionality of the data and many factors of variation involved. In this work, we investigate the task of single-view 3D object reconstruction from a learning agent's perspective. We formulate the learning process as an interaction between 3D and 2D representations and propose an encoder-decoder network with a novel projection loss defined by the perspective transformation. More importantly, the projection loss enables the unsupervised learning using 2D observation without explicit 3D supervision. We demonstrate the ability of the model in generating 3D volume from a single 2D image with three sets of experiments: (1) learning from single-class objects; (2) learning from multi-class objects and (3) testing on novel object classes. Results show superior performance and better generalization ability for 3D object reconstruction when the projection loss is involved.", "Object reconstruction from a single image - in the wild - is a problem where we can make progress and get meaningful results today. This is the main message of this paper, which introduces an automated pipeline with pixels as inputs and 3D surfaces of various rigid categories as outputs in images of realistic scenes. At the core of our approach are deformable 3D models that can be learned from 2D annotations available in existing object detection datasets, that can be driven by noisy automatic object segmentations and which we complement with a bottom-up module for recovering high-frequency shape details. We perform a comprehensive quantitative analysis and ablation study of our approach using the recently introduced PASCAL 3D+ dataset and show very encouraging automatic reconstructions on PASCAL VOC.", "", "What is a good vector representation of an object? We believe that it should be generative in 3D, in the sense that it can produce new 3D objects; as well as be predictable from 2D, in the sense that it can be perceived from 2D images. We propose a novel architecture, called the TL-embedding network, to learn an embedding space with these properties. The network consists of two components: (a) an autoencoder that ensures the representation is generative; and (b) a convolutional network that ensures the representation is predictable. This enables tackling a number of tasks including voxel prediction from 2D images and 3D model retrieval. Extensive experimental analysis demonstrates the usefulness and versatility of this embedding.", "3D object reconstruction from a single image is a highly under-determined problem, requiring strong prior knowledge of plausible 3D shapes. This introduces challenge for learning-based approaches, as 3D object annotations in real images are scarce. Previous work chose to train on synthetic data with ground truth 3D information, but suffered from the domain adaptation issue when tested on real data. In this work, we propose an end-to-end trainable framework, sequentially estimating 2.5D sketches and 3D object shapes. Our disentangled, two-step formulation has three advantages. First, compared to full 3D shape, 2.5D sketches are much easier to be recovered from a 2D image, and to transfer from synthetic to real data. Second, for 3D reconstruction from the 2.5D sketches, we can easily transfer the learned model on synthetic data to real images, as rendered 2.5D sketches are invariant to object appearance variations in real images, including lighting, texture, etc. This further relieves the domain adaptation problem. Third, we derive differentiable projective functions from 3D shape to 2.5D sketches, making the framework end-to-end trainable on real images, requiring no real-image annotations. Our framework achieves state-of-the-art performance on 3D shape reconstruction.", "We study the notion of consistency between a 3D shape and a 2D observation and propose a differentiable formulation which allows computing gradients of the 3D shape given an observation from an arbitrary view. We do so by reformulating view consistency using a differentiable ray consistency (DRC) term. We show that this formulation can be incorporated in a learning framework to leverage different types of multi-view observations e.g. foreground masks, depth, color images, semantics etc. as supervision for learning single-view 3D prediction. We present empirical analysis of our technique in a controlled setting. We also show that this approach allows us to improve over existing techniques for single-view reconstruction of objects from the PASCAL VOC dataset.", "This paper presents a quantitative comparison of several multi-view stereo reconstruction algorithms. Until now, the lack of suitable calibrated multi-view image datasets with known ground truth (3D shape models) has prevented such direct comparisons. In this paper, we first survey multi-view stereo algorithms and compare them qualitatively using a taxonomy that differentiates their key properties. We then describe our process for acquiring and calibrating multiview image datasets with high-accuracy ground truth and introduce our evaluation methodology. Finally, we present the results of our quantitative comparison of state-of-the-art multi-view stereo reconstruction algorithms on six benchmark datasets. The datasets, evaluation details, and instructions for submitting new models are available online at http: vision.middlebury.edu mview.", "It is expensive to label images with 3D structure or precise camera pose. Yet, this is precisely the kind of annotation required to train single-view 3D reconstruction models. In contrast, unlabeled images or images with just category labels are easy to acquire, but few current models can use this weak supervision. We present a unified framework that can combine both types of supervision: a small amount of camera pose annotations are used to enforce pose-invariance and view-point consistency, and unlabeled images combined with an adversarial loss are used to enforce the realism of rendered, generated models. We use this unified framework to measure the impact of each form of supervision in three paradigms: semi-supervised, multi-task, and transfer learning. We show that with a combination of these ideas, we can train single-view reconstruction models that improve up to 7 points in performance (AP) when using only 1 pose annotated training data.", "Generation of 3D data by deep neural network has been attracting increasing attention in the research community. The majority of extant works resort to regular representations such as volumetric grids or collection of images, however, these representations obscure the natural invariance of 3D shapes under geometric transformations, and also suffer from a number of other issues. In this paper we address the problem of 3D reconstruction from a single image, generating a straight-forward form of output – point cloud coordinates. Along with this problem arises a unique and interesting issue, that the groundtruth shape for an input image may be ambiguous. Driven by this unorthordox output form and the inherent ambiguity in groundtruth, we design architecture, loss function and learning paradigm that are novel and effective. Our final solution is a conditional shape sampler, capable of predicting multiple plausible 3D point clouds from an input image. In experiments not only can our system outperform state-of-the-art methods on single image based 3D reconstruction benchmarks, but it also shows strong performance for 3D shape completion and promising ability in making multiple plausible predictions.", "A key goal of computer vision is to recover the underlying 3D structure that gives rise to 2D observations of the world. If endowed with 3D understanding, agents can abstract away from the complexity of the rendering process to form stable, disentangled representations of scene elements. In this paper we learn strong deep generative models of 3D structures, and recover these structures from 2D images via probabilistic inference. We demonstrate high-quality samples and report log-likelihoods on several datasets, including ShapeNet, and establish the first benchmarks in the literature. We also show how these models and their inference networks can be trained jointly, end-to-end, and directly from 2D images without any use of ground-truth 3D labels. This demonstrates for the first time the feasibility of learning to infer 3D representations of the world in a purely unsupervised manner.", "Inspired by the recent success of methods that employ shape priors to achieve robust 3D reconstructions, we propose a novel recurrent neural network architecture that we call the 3D Recurrent Reconstruction Neural Network (3D-R2N2). The network learns a mapping from images of objects to their underlying 3D shapes from a large collection of synthetic data [13]. Our network takes in one or more images of an object instance from arbitrary viewpoints and outputs a reconstruction of the object in the form of a 3D occupancy grid. Unlike most of the previous works, our network does not require any image annotations or object class labels for training or testing. Our extensive experimental analysis shows that our reconstruction framework (i) outperforms the state-of-the-art methods for single view reconstruction, and (ii) enables the 3D reconstruction of objects in situations when traditional SFM SLAM methods fail (because of lack of texture and or wide baseline).", "This paper introduces a new probabilistic framework for Space Carving. In this framework each voxel is assigned a probability, which is computed by comparing the likelihoods for the voxel existing and not existing. This new framework avoids many of the difficulties associated with the original Space Carving algorithm. Specifically, it does not need a global threshold parameter, and it guarantees that no holes will be carved in the model. This paper also proposes that a voxel-based thick texture is a realistic and efficient representation for scenes which contain dominant planes. The algorithm is tested using both real and synthetic data, and both qualitative and quantitative results are presented.", "Many algorithms for both identifying and reconstructing a 3-D object are based on the 2-D silhouettes of the object. In general, identifying a nonconvex object using a silhouette-based approach implies neglecting some features of its surface as identification clues. The same features cannot be reconstructed by volume intersection techniques using multiple silhouettes of the object. This paper addresses the problem of finding which parts of a nonconvex object are relevant for silhouette-based image understanding. For this purpose, the geometric concept of visual hull of a 3-D object is introduced. This is the closest approximation of object S that can be obtained with the volume intersection approach; it is the maximal object silhouette-equivalent to S, i.e., which can be substituted for S without affecting any silhouette. Only the parts of the surface of S that also lie on the surface of the visual hull can be reconstructed or identified using silhouette-based algorithms. The visual hull depends not only on the object but also on the region allowed to the viewpoint. Two main viewing regions result in the external and internal visual hull. In the former case the viewing region is related to the convex hull of S, in the latter it is bounded by S. The internal visual hull also admits an interpretation not related to silhouettes. Algorithms for computing visual hulls are presented and their complexity analyzed. In general, the visual hull of a 3-D planar face object turns out to be bounded by planar and curved patches. >", "This paper examines the problem of reconstructing a voxelized representation of 3D space from a series of images. An iterative algorithm is used to find the scene model which jointly explains all the observed images by determining which region of space is responsible for each of the observations. The current approach formulates the problem as one of optimization over estimates of these responsibilities. The process converges to a distribution of responsibility which accurately reflects the constraints provided by the observations, the positions and shape of both solid and transparent objects, and the uncertainty which remains. Reconstruction is robust, and gracefully represents regions of space in which there is little certainty about the exact structure due to limited, non-existent, or contradicting data. Rendered images of voxel spaces recovered from synthetic and real observation images are shown.", "3D shape is a crucial but heavily underutilized cue in today's computer vision systems, mostly due to the lack of a good generic shape representation. With the recent availability of inexpensive 2.5D depth sensors (e.g. Microsoft Kinect), it is becoming increasingly important to have a powerful 3D shape representation in the loop. Apart from category recognition, recovering full 3D shapes from view-based 2.5D depth maps is also a critical part of visual understanding. To this end, we propose to represent a geometric 3D shape as a probability distribution of binary variables on a 3D voxel grid, using a Convolutional Deep Belief Network. Our model, 3D ShapeNets, learns the distribution of complex 3D shapes across different object categories and arbitrary poses from raw CAD data, and discovers hierarchical compositional part representation automatically. It naturally supports joint object recognition and shape completion from 2.5D depth maps, and it enables active object recognition through view planning. To train our 3D deep learning model, we construct ModelNet - a large-scale 3D CAD model dataset. Extensive experiments show that our 3D deep representation enables significant performance improvement over the-state-of-the-arts in a variety of tasks.", "We address the problem of populating object category detection datasets with dense, per-object 3D reconstructions, bootstrapped from class labels, ground truth figure-ground segmentations and a small set of keypoint annotations. Our proposed algorithm first estimates camera viewpoint using rigid structure-from-motion, then reconstructs object shapes by optimizing over visual hull proposals guided by loose within-class shape similarity assumptions. The visual hull sampling process attempts to intersect an object's projection cone with the cones of minimal subsets of other similar objects among those pictured from certain vantage points. We show that our method is able to produce convincing per-object 3D reconstructions on one of the most challenging existing object-category detection datasets, PASCAL VOC. Our results may re-stimulate once popular geometry-oriented model-based recognition approaches." ] }
1812.01742	2969824030	Single-view 3D shape reconstruction is an important but challenging problem, mainly for two reasons. First, as shape annotation is very expensive to acquire, current methods rely on synthetic data, in which ground-truth 3D annotation is easy to obtain. However, this results in domain adaptation problem when applied to natural images. The second challenge is that there are multiple shapes that can explain a given 2D image. In this paper, we propose a framework to improve over these challenges using adversarial training. On one hand, we impose domain confusion between natural and synthetic image representations to reduce the distribution gap. On the other hand, we impose the reconstruction to be realistic' by forcing it to lie on a (learned) manifold of realistic object shapes. Our experiments show that these constraints improve performance by a large margin over baseline reconstruction models. We achieve results competitive with the state of the art with a much simpler architecture.	These approaches, based on deep learning techniques, usually share a similar (high-level) architecture: an encoder that maps 2D (rendered) images into a latent representation and a decoder that maps this representation into a 3D object. They tend to differ in the way 3D world constraint are imposed. For instance, @cite_44 @cite_35 @cite_35 @cite_48 @cite_26 @cite_39 @cite_31 force multiview consistency to learn the 3D representation, while @cite_20 @cite_4 @cite_42 leverage keypoints and silhouette annotations. Other approaches @cite_28 @cite_8 leverage 2.5 sketches (surface normals, depth and silhouette) information to improve prediction.	{ "cite_N": [ "@cite_35", "@cite_26", "@cite_4", "@cite_8", "@cite_28", "@cite_48", "@cite_42", "@cite_39", "@cite_44", "@cite_31", "@cite_20" ], "mid": [ "2950701417", "2619556892", "1893912098", "2890382763", "2963433432", "2963739349", "2963850211", "2963730200", "2342277278", "", "1977792424" ], "abstract": [ "Understanding the 3D world is a fundamental problem in computer vision. However, learning a good representation of 3D objects is still an open problem due to the high dimensionality of the data and many factors of variation involved. In this work, we investigate the task of single-view 3D object reconstruction from a learning agent's perspective. We formulate the learning process as an interaction between 3D and 2D representations and propose an encoder-decoder network with a novel projection loss defined by the perspective transformation. More importantly, the projection loss enables the unsupervised learning using 2D observation without explicit 3D supervision. We demonstrate the ability of the model in generating 3D volume from a single 2D image with three sets of experiments: (1) learning from single-class objects; (2) learning from multi-class objects and (3) testing on novel object classes. Results show superior performance and better generalization ability for 3D object reconstruction when the projection loss is involved.", "", "Object reconstruction from a single image - in the wild - is a problem where we can make progress and get meaningful results today. This is the main message of this paper, which introduces an automated pipeline with pixels as inputs and 3D surfaces of various rigid categories as outputs in images of realistic scenes. At the core of our approach are deformable 3D models that can be learned from 2D annotations available in existing object detection datasets, that can be driven by noisy automatic object segmentations and which we complement with a bottom-up module for recovering high-frequency shape details. We perform a comprehensive quantitative analysis and ablation study of our approach using the recently introduced PASCAL 3D+ dataset and show very encouraging automatic reconstructions on PASCAL VOC.", "The problem of single-view 3D shape completion or reconstruction is challenging, because among the many possible shapes that explain an observation, most are implausible and do not correspond to natural objects. Recent research in the field has tackled this problem by exploiting the expressiveness of deep convolutional networks. In fact, there is another level of ambiguity that is often overlooked: among plausible shapes, there are still multiple shapes that fit the 2D image equally well; i.e., the ground truth shape is non-deterministic given a single-view input. Existing fully supervised approaches fail to address this issue, and often produce blurry mean shapes with smooth surfaces but no fine details. In this paper, we propose ShapeHD, pushing the limit of single-view shape completion and reconstruction by integrating deep generative models with adversarially learned shape priors. The learned priors serve as a regularizer, penalizing the model only if its output is unrealistic, not if it deviates from the ground truth. Our design thus overcomes both levels of ambiguity aforementioned. Experiments demonstrate that ShapeHD outperforms state of the art by a large margin in both shape completion and shape reconstruction on multiple real datasets.", "3D object reconstruction from a single image is a highly under-determined problem, requiring strong prior knowledge of plausible 3D shapes. This introduces challenge for learning-based approaches, as 3D object annotations in real images are scarce. Previous work chose to train on synthetic data with ground truth 3D information, but suffered from the domain adaptation issue when tested on real data. In this work, we propose an end-to-end trainable framework, sequentially estimating 2.5D sketches and 3D object shapes. Our disentangled, two-step formulation has three advantages. First, compared to full 3D shape, 2.5D sketches are much easier to be recovered from a 2D image, and to transfer from synthetic to real data. Second, for 3D reconstruction from the 2.5D sketches, we can easily transfer the learned model on synthetic data to real images, as rendered 2.5D sketches are invariant to object appearance variations in real images, including lighting, texture, etc. This further relieves the domain adaptation problem. Third, we derive differentiable projective functions from 3D shape to 2.5D sketches, making the framework end-to-end trainable on real images, requiring no real-image annotations. Our framework achieves state-of-the-art performance on 3D shape reconstruction.", "We study the notion of consistency between a 3D shape and a 2D observation and propose a differentiable formulation which allows computing gradients of the 3D shape given an observation from an arbitrary view. We do so by reformulating view consistency using a differentiable ray consistency (DRC) term. We show that this formulation can be incorporated in a learning framework to leverage different types of multi-view observations e.g. foreground masks, depth, color images, semantics etc. as supervision for learning single-view 3D prediction. We present empirical analysis of our technique in a controlled setting. We also show that this approach allows us to improve over existing techniques for single-view reconstruction of objects from the PASCAL VOC dataset.", "We present a learning framework for recovering the 3D shape, camera, and texture of an object from a single image. The shape is represented as a deformable 3D mesh model of an object category where a shape is parameterized by a learned mean shape and per-instance predicted deformation. Our approach allows leveraging an annotated image collection for training, where the deformable model and the 3D prediction mechanism are learned without relying on ground-truth 3D or multi-view supervision. Our representation enables us to go beyond existing 3D prediction approaches by incorporating texture inference as prediction of an image in a canonical appearance space. Additionally, we show that semantic keypoints can be easily associated with the predicted shapes. We present qualitative and quantitative results of our approach on CUB and PASCAL3D datasets and show that we can learn to predict diverse shapes and textures across objects using only annotated image collections. The project website can be found at https: akanazawa.github.io cmr .", "A key goal of computer vision is to recover the underlying 3D structure that gives rise to 2D observations of the world. If endowed with 3D understanding, agents can abstract away from the complexity of the rendering process to form stable, disentangled representations of scene elements. In this paper we learn strong deep generative models of 3D structures, and recover these structures from 2D images via probabilistic inference. We demonstrate high-quality samples and report log-likelihoods on several datasets, including ShapeNet, and establish the first benchmarks in the literature. We also show how these models and their inference networks can be trained jointly, end-to-end, and directly from 2D images without any use of ground-truth 3D labels. This demonstrates for the first time the feasibility of learning to infer 3D representations of the world in a purely unsupervised manner.", "Inspired by the recent success of methods that employ shape priors to achieve robust 3D reconstructions, we propose a novel recurrent neural network architecture that we call the 3D Recurrent Reconstruction Neural Network (3D-R2N2). The network learns a mapping from images of objects to their underlying 3D shapes from a large collection of synthetic data [13]. Our network takes in one or more images of an object instance from arbitrary viewpoints and outputs a reconstruction of the object in the form of a 3D occupancy grid. Unlike most of the previous works, our network does not require any image annotations or object class labels for training or testing. Our extensive experimental analysis shows that our reconstruction framework (i) outperforms the state-of-the-art methods for single view reconstruction, and (ii) enables the 3D reconstruction of objects in situations when traditional SFM SLAM methods fail (because of lack of texture and or wide baseline).", "", "We address the problem of populating object category detection datasets with dense, per-object 3D reconstructions, bootstrapped from class labels, ground truth figure-ground segmentations and a small set of keypoint annotations. Our proposed algorithm first estimates camera viewpoint using rigid structure-from-motion, then reconstructs object shapes by optimizing over visual hull proposals guided by loose within-class shape similarity assumptions. The visual hull sampling process attempts to intersect an object's projection cone with the cones of minimal subsets of other similar objects among those pictured from certain vantage points. We show that our method is able to produce convincing per-object 3D reconstructions on one of the most challenging existing object-category detection datasets, PASCAL VOC. Our results may re-stimulate once popular geometry-oriented model-based recognition approaches." ] }
1812.01742	2969824030	Single-view 3D shape reconstruction is an important but challenging problem, mainly for two reasons. First, as shape annotation is very expensive to acquire, current methods rely on synthetic data, in which ground-truth 3D annotation is easy to obtain. However, this results in domain adaptation problem when applied to natural images. The second challenge is that there are multiple shapes that can explain a given 2D image. In this paper, we propose a framework to improve over these challenges using adversarial training. On one hand, we impose domain confusion between natural and synthetic image representations to reduce the distribution gap. On the other hand, we impose the reconstruction to be realistic' by forcing it to lie on a (learned) manifold of realistic object shapes. Our experiments show that these constraints improve performance by a large margin over baseline reconstruction models. We achieve results competitive with the state of the art with a much simpler architecture.	The difficulty to acquire 3D annotations for natural images forces reconstruction models to learn form rendered images. It is well known in the literature @cite_6 @cite_24 that the performance of a model drops if applied in data coming from a distribution different from the one used during training. Ganin @cite_23 deal with this issue by forcing domain confusion (between two domains) through an adversarial objective. Many works have been dealing with domain adaptation from synthetic to real for image classification @cite_22 @cite_34 @cite_46 @cite_16 . In this work, we borrow ideas from domain adaptation literature to impose domain confusion in a similar way as these previous work. We consider, however, the more challenging problem of 3D reconstruction instead of classification.	{ "cite_N": [ "@cite_22", "@cite_6", "@cite_16", "@cite_24", "@cite_23", "@cite_46", "@cite_34" ], "mid": [ "2964139811", "2031342017", "2798593490", "2756073160", "1731081199", "2963993484", "2962687275" ], "abstract": [ "We present a method for transferring neural representations from label-rich source domains to unlabeled target domains. Recent adversarial methods proposed for this task learn to align features across domains by fooling a special domain critic network. However, a drawback of this approach is that the critic simply labels the generated features as in-domain or not, without considering the boundaries between classes. This can lead to ambiguous features being generated near class boundaries, reducing target classification accuracy. We propose a novel approach, Adversarial Dropout Regularization (ADR), to encourage the generator to output more discriminative features for the target domain. Our key idea is to replace the critic with one that detects non-discriminative features, using dropout on the classifier network. The generator then learns to avoid these areas of the feature space and thus creates better features. We apply our ADR approach to the problem of unsupervised domain adaptation for image classification and semantic segmentation tasks, and demonstrate significant improvement over the state of the art. We also show that our approach can be used to train Generative Adversarial Networks for semi-supervised learning.", "Datasets are an integral part of contemporary object recognition research. They have been the chief reason for the considerable progress in the field, not just as source of large amounts of training data, but also as means of measuring and comparing performance of competing algorithms. At the same time, datasets have often been blamed for narrowing the focus of object recognition research, reducing it to a single benchmark performance number. Indeed, some datasets, that started out as data capture efforts aimed at representing the visual world, have become closed worlds unto themselves (e.g. the Corel world, the Caltech-101 world, the PASCAL VOC world). With the focus on beating the latest benchmark numbers on the latest dataset, have we perhaps lost sight of the original purpose? The goal of this paper is to take stock of the current state of recognition datasets. We present a comparison study using a set of popular datasets, evaluated based on a number of criteria including: relative data bias, cross-dataset generalization, effects of closed-world assumption, and sample value. The experimental results, some rather surprising, suggest directions that can improve dataset collection as well as algorithm evaluation protocols. But more broadly, the hope is to stimulate discussion in the community regarding this very important, but largely neglected issue.", "Numerous algorithms have been proposed for transferring knowledge from a label-rich domain (source) to a label-scarce domain (target). Most of them are proposed for closed-set scenario, where the source and the target domain completely share the class of their samples. However, in practice, a target domain can contain samples of classes that are not shared by the source domain. We call such classes the “unknown class” and algorithms that work well in the open set situation are very practical. However, most existing distribution matching methods for domain adaptation do not work well in this setting because unknown target samples should not be aligned with the source. In this paper, we propose a method for an open set domain adaptation scenario, which utilizes adversarial training. This approach allows to extract features that separate unknown target from known target samples. During training, we assign two options to the feature generator: aligning target samples with source known ones or rejecting them as unknown target ones. Our method was extensively evaluated and outperformed other methods with a large margin in most settings.", "The aim of this chapter is to give an overview of domain adaptation and transfer learning with a specific view to visual applications. After a general motivation, we first position domain adaptation in the more general transfer learning problem. Second, we try to address and analyze briefly the state-of-the-art methods for different types of scenarios, first describing the historical shallow methods, addressing both the homogeneous and heterogeneous domain adaptation methods. Third, we discuss the effect of the success of deep convolutional architectures which led to the new type of domain adaptation methods that integrate the adaptation within the deep architecture. Fourth, we review DA methods that go beyond image categorization, such as object detection, image segmentation, video analyses or learning visual attributes. We conclude the chapter with a section where we relate domain adaptation to other machine learning solutions.", "We introduce a new representation learning approach for domain adaptation, in which data at training and test time come from similar but different distributions. Our approach is directly inspired by the theory on domain adaptation suggesting that, for effective domain transfer to be achieved, predictions must be made based on features that cannot discriminate between the training (source) and test (target) domains. The approach implements this idea in the context of neural network architectures that are trained on labeled data from the source domain and unlabeled data from the target domain (no labeled target-domain data is necessary). As the training progresses, the approach promotes the emergence of features that are (i) discriminative for the main learning task on the source domain and (ii) indiscriminate with respect to the shift between the domains. We show that this adaptation behaviour can be achieved in almost any feed-forward model by augmenting it with few standard layers and a new gradient reversal layer. The resulting augmented architecture can be trained using standard backpropagation and stochastic gradient descent, and can thus be implemented with little effort using any of the deep learning packages. We demonstrate the success of our approach for two distinct classification problems (document sentiment analysis and image classification), where state-of-the-art domain adaptation performance on standard benchmarks is achieved. We also validate the approach for descriptor learning task in the context of person re-identification application.", "The objective of unsupervised domain adaptation is to leverage features from a labeled source domain and learn a classifier for an unlabeled target domain, with a similar but different data distribution. Most deep learning approaches to domain adaptation consist of two steps: (i) learn features that preserve a low risk on labeled samples (source domain) and (ii) make the features from both domains to be as indistinguishable as possible, so that a classifier trained on the source can also be applied on the target domain. In general, the classifiers in step (i) consist of fully-connected layers applied directly on the indistinguishable features learned in (ii). In this paper, we propose a different way to do the classification, using similarity learning. The proposed method learns a pairwise similarity function in which classification can be performed by computing similarity between prototype representations of each category. The domain-invariant features and the categorical prototype representations are learned jointly and in an end-to-end fashion. At inference time, images from the target domain are compared to the prototypes and the label associated with the one that best matches the image is outputed. The approach is simple, scalable and effective. We show that our model achieves state-of-the-art performance in different unsupervised domain adaptation scenarios.", "In this work, we present a method for unsupervised domain adaptation. Many adversarial learning methods train domain classifier networks to distinguish the features as either a source or target and train a feature generator network to mimic the discriminator. Two problems exist with these methods. First, the domain classifier only tries to distinguish the features as a source or target and thus does not consider task-specific decision boundaries between classes. Therefore, a trained generator can generate ambiguous features near class boundaries. Second, these methods aim to completely match the feature distributions between different domains, which is difficult because of each domain's characteristics. To solve these problems, we introduce a new approach that attempts to align distributions of source and target by utilizing the task-specific decision boundaries. We propose to maximize the discrepancy between two classifiers' outputs to detect target samples that are far from the support of the source. A feature generator learns to generate target features near the support to minimize the discrepancy. Our method outperforms other methods on several datasets of image classification and semantic segmentation. The codes are available at https: github.com mil-tokyo MCD_DA" ] }
1812.01742	2969824030	Single-view 3D shape reconstruction is an important but challenging problem, mainly for two reasons. First, as shape annotation is very expensive to acquire, current methods rely on synthetic data, in which ground-truth 3D annotation is easy to obtain. However, this results in domain adaptation problem when applied to natural images. The second challenge is that there are multiple shapes that can explain a given 2D image. In this paper, we propose a framework to improve over these challenges using adversarial training. On one hand, we impose domain confusion between natural and synthetic image representations to reduce the distribution gap. On the other hand, we impose the reconstruction to be realistic' by forcing it to lie on a (learned) manifold of realistic object shapes. Our experiments show that these constraints improve performance by a large margin over baseline reconstruction models. We achieve results competitive with the state of the art with a much simpler architecture.	Reconstruction of 3D structure from single-view images requires strong priors about object's shape. Many works focus on better capturing the manifold of realistic shapes. Non-deep approaches had focus on low-dimensional parametric models @cite_43 @cite_4 . @cite_36 @cite_7 use CNNs to learn a common embedding space for 2D rendered images and 3D shapes. Other methods rely on generative modeling to learn shape prior, , @cite_15 use deep belief nets to model 3D representations, @cite_39 @cite_49 @cite_52 consider variants of variational autoencoders and @cite_37 use a variant of GANs @cite_47 to capture the manifold of shapes.	{ "cite_N": [ "@cite_37", "@cite_47", "@cite_4", "@cite_7", "@cite_36", "@cite_52", "@cite_39", "@cite_43", "@cite_49", "@cite_15" ], "mid": [ "2546066744", "2099471712", "1893912098", "2083163329", "2964137676", "", "2963730200", "2237250383", "2116546169", "1920022804" ], "abstract": [ "We study the problem of 3D object generation. We propose a novel framework, namely 3D Generative Adversarial Network (3D-GAN), which generates 3D objects from a probabilistic space by leveraging recent advances in volumetric convolutional networks and generative adversarial nets. The benefits of our model are three-fold: first, the use of an adversarial criterion, instead of traditional heuristic criteria, enables the generator to capture object structure implicitly and to synthesize high-quality 3D objects; second, the generator establishes a mapping from a low-dimensional probabilistic space to the space of 3D objects, so that we can sample objects without a reference image or CAD models, and explore the 3D object manifold; third, the adversarial discriminator provides a powerful 3D shape descriptor which, learned without supervision, has wide applications in 3D object recognition. Experiments demonstrate that our method generates high-quality 3D objects, and our unsupervisedly learned features achieve impressive performance on 3D object recognition, comparable with those of supervised learning methods.", "We propose a new framework for estimating generative models via an adversarial process, in which we simultaneously train two models: a generative model G that captures the data distribution, and a discriminative model D that estimates the probability that a sample came from the training data rather than G. The training procedure for G is to maximize the probability of D making a mistake. This framework corresponds to a minimax two-player game. In the space of arbitrary functions G and D, a unique solution exists, with G recovering the training data distribution and D equal to ½ everywhere. In the case where G and D are defined by multilayer perceptrons, the entire system can be trained with backpropagation. There is no need for any Markov chains or unrolled approximate inference networks during either training or generation of samples. Experiments demonstrate the potential of the framework through qualitative and quantitative evaluation of the generated samples.", "Object reconstruction from a single image - in the wild - is a problem where we can make progress and get meaningful results today. This is the main message of this paper, which introduces an automated pipeline with pixels as inputs and 3D surfaces of various rigid categories as outputs in images of realistic scenes. At the core of our approach are deformable 3D models that can be learned from 2D annotations available in existing object detection datasets, that can be driven by noisy automatic object segmentations and which we complement with a bottom-up module for recovering high-frequency shape details. We perform a comprehensive quantitative analysis and ablation study of our approach using the recently introduced PASCAL 3D+ dataset and show very encouraging automatic reconstructions on PASCAL VOC.", "Both 3D models and 2D images contain a wealth of information about everyday objects in our environment. However, it is difficult to semantically link together these two media forms, even when they feature identical or very similar objects. We propose a joint embedding space populated by both 3D shapes and 2D images of objects, where the distances between embedded entities reflect similarity between the underlying objects. This joint embedding space facilitates comparison between entities of either form, and allows for cross-modality retrieval. We construct the embedding space using 3D shape similarity measure, as 3D shapes are more pure and complete than their appearance in images, leading to more robust distance metrics. We then employ a Convolutional Neural Network (CNN) to \"purify\" images by muting distracting factors. The CNN is trained to map an image to a point in the embedding space, so that it is close to a point attributed to a 3D model of a similar object to the one depicted in the image. This purifying capability of the CNN is accomplished with the help of a large amount of training data consisting of images synthesized from 3D shapes. Our joint embedding allows cross-view image retrieval, image-based shape retrieval, as well as shape-based image retrieval. We evaluate our method on these retrieval tasks and show that it consistently out-performs state-of-the-art methods, and demonstrate the usability of a joint embedding in a number of additional applications.", "What is a good vector representation of an object? We believe that it should be generative in 3D, in the sense that it can produce new 3D objects; as well as be predictable from 2D, in the sense that it can be perceived from 2D images. We propose a novel architecture, called the TL-embedding network, to learn an embedding space with these properties. The network consists of two components: (a) an autoencoder that ensures the representation is generative; and (b) a convolutional network that ensures the representation is predictable. This enables tackling a number of tasks including voxel prediction from 2D images and 3D model retrieval. Extensive experimental analysis demonstrates the usefulness and versatility of this embedding.", "", "A key goal of computer vision is to recover the underlying 3D structure that gives rise to 2D observations of the world. If endowed with 3D understanding, agents can abstract away from the complexity of the rendering process to form stable, disentangled representations of scene elements. In this paper we learn strong deep generative models of 3D structures, and recover these structures from 2D images via probabilistic inference. We demonstrate high-quality samples and report log-likelihoods on several datasets, including ShapeNet, and establish the first benchmarks in the literature. We also show how these models and their inference networks can be trained jointly, end-to-end, and directly from 2D images without any use of ground-truth 3D labels. This demonstrates for the first time the feasibility of learning to infer 3D representations of the world in a purely unsupervised manner.", "In this paper, a new technique for modeling textured 3D faces is introduced. 3D faces can either be generated automatically from one or more photographs, or modeled directly through an intuitive user interface. Users are assisted in two key problems of computer aided face modeling. First, new face images or new 3D face models can be registered automatically by computing dense one-to-one correspondence to an internal face model. Second, the approach regulates the naturalness of modeled faces avoiding faces with an “unlikely” appearance. Starting from an example set of 3D face models, we derive a morphable face model by transforming the shape and texture of the examples into a vector space representation. New faces and expressions can be modeled by forming linear combinations of the prototypes. Shape and texture constraints derived from the statistics of our example faces are used to guide manual modeling or automated matching algorithms. We show 3D face reconstructions from single images and their applications for photo-realistic image manipulations. We also demonstrate face manipulations according to complex parameters such as gender, fullness of a face or its distinctiveness.", "This paper introduces a new probabilistic framework for Space Carving. In this framework each voxel is assigned a probability, which is computed by comparing the likelihoods for the voxel existing and not existing. This new framework avoids many of the difficulties associated with the original Space Carving algorithm. Specifically, it does not need a global threshold parameter, and it guarantees that no holes will be carved in the model. This paper also proposes that a voxel-based thick texture is a realistic and efficient representation for scenes which contain dominant planes. The algorithm is tested using both real and synthetic data, and both qualitative and quantitative results are presented.", "3D shape is a crucial but heavily underutilized cue in today's computer vision systems, mostly due to the lack of a good generic shape representation. With the recent availability of inexpensive 2.5D depth sensors (e.g. Microsoft Kinect), it is becoming increasingly important to have a powerful 3D shape representation in the loop. Apart from category recognition, recovering full 3D shapes from view-based 2.5D depth maps is also a critical part of visual understanding. To this end, we propose to represent a geometric 3D shape as a probability distribution of binary variables on a 3D voxel grid, using a Convolutional Deep Belief Network. Our model, 3D ShapeNets, learns the distribution of complex 3D shapes across different object categories and arbitrary poses from raw CAD data, and discovers hierarchical compositional part representation automatically. It naturally supports joint object recognition and shape completion from 2.5D depth maps, and it enables active object recognition through view planning. To train our 3D deep learning model, we construct ModelNet - a large-scale 3D CAD model dataset. Extensive experiments show that our 3D deep representation enables significant performance improvement over the-state-of-the-arts in a variety of tasks." ] }
1812.01742	2969824030	Single-view 3D shape reconstruction is an important but challenging problem, mainly for two reasons. First, as shape annotation is very expensive to acquire, current methods rely on synthetic data, in which ground-truth 3D annotation is easy to obtain. However, this results in domain adaptation problem when applied to natural images. The second challenge is that there are multiple shapes that can explain a given 2D image. In this paper, we propose a framework to improve over these challenges using adversarial training. On one hand, we impose domain confusion between natural and synthetic image representations to reduce the distribution gap. On the other hand, we impose the reconstruction to be realistic' by forcing it to lie on a (learned) manifold of realistic object shapes. Our experiments show that these constraints improve performance by a large margin over baseline reconstruction models. We achieve results competitive with the state of the art with a much simpler architecture.	A few works use adversarial training for singe-view 3D reconstruction. Gwak @cite_26 use GANs to model 2D projections instead of 3D shapes. More similar to our work, Wu, Zhang @cite_8 use adversarial training techniques to impose reconstructions to look more natural. They use the discriminator of a pre-trained 3D GAN @cite_37 to determine whether a shape is realistic. This approach is similar in principle to one of our contributions. It is, however, implemented in very different way. The input to the discriminator is a high dimensional 3D shape, which makes the training to be very unstable. In our method, the input is a single vector in a low-dimensional space.	{ "cite_N": [ "@cite_37", "@cite_26", "@cite_8" ], "mid": [ "2546066744", "2619556892", "2890382763" ], "abstract": [ "We study the problem of 3D object generation. We propose a novel framework, namely 3D Generative Adversarial Network (3D-GAN), which generates 3D objects from a probabilistic space by leveraging recent advances in volumetric convolutional networks and generative adversarial nets. The benefits of our model are three-fold: first, the use of an adversarial criterion, instead of traditional heuristic criteria, enables the generator to capture object structure implicitly and to synthesize high-quality 3D objects; second, the generator establishes a mapping from a low-dimensional probabilistic space to the space of 3D objects, so that we can sample objects without a reference image or CAD models, and explore the 3D object manifold; third, the adversarial discriminator provides a powerful 3D shape descriptor which, learned without supervision, has wide applications in 3D object recognition. Experiments demonstrate that our method generates high-quality 3D objects, and our unsupervisedly learned features achieve impressive performance on 3D object recognition, comparable with those of supervised learning methods.", "", "The problem of single-view 3D shape completion or reconstruction is challenging, because among the many possible shapes that explain an observation, most are implausible and do not correspond to natural objects. Recent research in the field has tackled this problem by exploiting the expressiveness of deep convolutional networks. In fact, there is another level of ambiguity that is often overlooked: among plausible shapes, there are still multiple shapes that fit the 2D image equally well; i.e., the ground truth shape is non-deterministic given a single-view input. Existing fully supervised approaches fail to address this issue, and often produce blurry mean shapes with smooth surfaces but no fine details. In this paper, we propose ShapeHD, pushing the limit of single-view shape completion and reconstruction by integrating deep generative models with adversarially learned shape priors. The learned priors serve as a regularizer, penalizing the model only if its output is unrealistic, not if it deviates from the ground truth. Our design thus overcomes both levels of ambiguity aforementioned. Experiments demonstrate that ShapeHD outperforms state of the art by a large margin in both shape completion and shape reconstruction on multiple real datasets." ] }
1812.01677	2902665249	With the aim of creating virtual cloth deformations more similar to real world clothing, we propose a new computational framework that recasts three dimensional cloth deformation as an RGB image in a two dimensional pattern space. Then a three dimensional animation of cloth is equivalent to a sequence of two dimensional RGB images, which in turn are driven choreographed via animation parameters such as joint angles. This allows us to leverage popular CNNs to learn cloth deformations in image space. The two dimensional cloth pixels are extended into the real world via standard body skinning techniques, after which the RGB values are interpreted as texture offsets and displacement maps. Notably, we illustrate that our approach does not require accurate unclothed body shapes or robust skinning techniques. Additionally, we discuss how standard image based techniques such as image partitioning for higher resolution, GANs for merging partitioned image regions back together, etc., can readily be incorporated into our framework.	Some of the aforementioned skinning type approaches to cloth and bodies learn from examples and therefore have procedural formulas and weights which often require optimization in order to define, but here we focus primarily on methods that use neural networks in a more data-driven as opposed to procedural fashion. While we utilize procedural methods for skinning the body mesh and subsequently finding our cloth pixel locations, we use data-driven networks to define the cloth deformations; errors in the procedural skinning are simply incorporated into the offset function used to subsequently reach the data. Several recent works used neural networks for learning 3D surface deformations for character rigs @cite_33 and cloth shapes @cite_3 @cite_13 @cite_48 . In particular, @cite_33 @cite_48 input pose parameters and output non-linear shape deformations of the skin cloth, both using a fully connected network with a few hidden layers to predict PCA coefficients. @cite_3 takes input images from single or multiple views and uses a convolutional network to predict 1000 PCA coefficients. @cite_13 takes a hybrid approach combining a statistical model for pose-based global deformation with a conditional generative adversarial network for adding details on normal maps to produce finer wrinkles.	{ "cite_N": [ "@cite_48", "@cite_13", "@cite_33", "@cite_3" ], "mid": [ "2887586359", "", "2810984579", "2607760177" ], "abstract": [ "Recent capture technologies and methods allow not only to retrieve 3D model sequence of moving people in clothing, but also to separate and extract the underlying body geometry and motion component and separate the clothing as a geometric layer. So far this clothing layer has only been used as raw offsets for individual applications such as retargeting a different body capture sequence with the clothing layer of another sequence, with limited scope, e.g. using identical or similar motions. The structured, semantics and motion-correlated nature of the information contained in this layer has yet to be fully understood and exploited. To this purpose we propose a comprehensive analysis of the statistics of this layer with a simple two-component model, based on PCA subspace reduction of the layer information on one hand, and a generic parameter regression model using neural networks on the other hand, designed to regress from any semantic parameter whose variation is observed in a training set, to the layer parameteriza-tion space. We show that this model not only allows to reproduce previous motion retargeting works, but generalizes the data generation capabilities of the method to other semantic parameters such as clothing variation and size, or physical material parameters with synthetically generated training sequence, paving the way for many kinds of capture data-driven creation and augmentation applications.", "", "Our method uses deep learning to approximate mesh deformations of film-quality characters, which allows the character rigs to run at interactive rates on consumer-quality devices.", "3D garment capture is an important component for various applications such as free-view point video, virtual avatars, online shopping, and virtual cloth fitting. Due to the complexity of the deformations, capturing 3D garment shapes requires controlled and specialized setups. A viable alternative is image-based garment capture. Capturing 3D garment shapes from a single image, however, is a challenging problem and the current solutions come with assumptions on the lighting, camera calibration, complexity of human or mannequin poses considered, and more importantly a stable physical state for the garment and the underlying human body. In addition, most of the works require manual interaction and exhibit high run-times. We propose a new technique that overcomes these limitations, making garment shape estimation from an image a practical approach for dynamic garment capture. Starting from synthetic garment shape data generated through physically based simulations from various human bodies in complex poses obtained through Mocap sequences, and rendered under varying camera positions and lighting conditions, our novel method learns a mapping from rendered garment images to the underlying 3D garment model. This is achieved by training Convolutional Neural Networks CNN-s to estimate 3D vertex displacements from a template mesh with a specialized loss function. We illustrate that this technique is able to recover the global shape of dynamic 3D garments from a single image under varying factors such as challenging human poses, self occlusions, various camera poses and lighting conditions, at interactive rates. Improvement is shown if more than one view is integrated. Additionally, we show applications of our method to videos." ] }
1812.01693	2903015761	The present online social media platform is afflicted with several issues, with hate speech being on the predominant forefront. The prevalence of online hate speech has fueled horrific real-world hate-crime such as the mass-genocide of Rohingya Muslims, communal violence in Colombo and the recent massacre in the Pittsburgh synagogue. Consequently, It is imperative to understand the diffusion of such hateful content in an online setting. We conduct the first study that analyses the flow and dynamics of posts generated by hateful and non-hateful users on Gab (gab.com) over a massive dataset of 341K users and 21M posts. Our observations confirms that hateful content diffuse farther, wider and faster and have a greater outreach than those of non-hateful users. A deeper inspection into the profiles and network of hateful and non-hateful users reveals that the former are more influential, popular and cohesive. Thus, our research explores the interesting facets of diffusion dynamics of hateful users and broadens our understanding of hate speech in the online world.	To the best of our knowledge there has not been any work that tries to study the diffusion of hate in online social media. However, there are several works that looks into diffusion in fake news @cite_18 @cite_8 @cite_4 , Linkedin @cite_35 , retweet cascade @cite_5 @cite_23 @cite_18 @cite_6 , rumours @cite_14 @cite_25 @cite_2 @cite_34 @cite_1 and Tumblr @cite_22 @cite_30 @cite_24 @cite_31 . @cite_6 perform large scale analysis of recurring cascades in Facebook. They observe that content virality is the main driver for recurrence. In @cite_1 , the authors perform a large scale analysis of Facebook and observe that selective exposure to content is the primary driver of content diffusion and generates the formation echo chambers.	{ "cite_N": [ "@cite_30", "@cite_35", "@cite_18", "@cite_14", "@cite_4", "@cite_22", "@cite_8", "@cite_1", "@cite_6", "@cite_24", "@cite_23", "@cite_2", "@cite_5", "@cite_31", "@cite_34", "@cite_25" ], "mid": [ "1982393014", "1417125139", "2790166049", "", "2783564496", "2735875961", "2604760541", "2232384272", "2257559988", "", "2178843456", "1638051351", "1996263819", "2288487960", "2164082612", "171462922" ], "abstract": [ "Tumblr, a microblogging platform and social media website, has been gaining popularity over the past few years. Despite its success, little has been studied on the human behavior and interaction on this platform. This is important as it sheds light on the driving force behind Tumblr's growth. In this work, we present a quantitative study of Tumblr based on the complete data coverage for four consecutive months consisting of 23.2 million users and 10.2 billion posts. We first explore various attributes of users, posts, and tags in detail and extract behavioral patterns based on the user generated content. We then construct a massive reblog network based on the primary user interactions on Tumblr and present findings on analyzing its topological structure and properties. Finally, we show substantial results on providing location-specific usage patterns from Tumblr, despite no built-in support for geo-tagging or user location functionality. Essentially this is done by conducting a large-scale user alignment with a different social media platform (e.g., Twitter) and subsequently propagating geo-information across platforms. To the best of our knowledge, this work is the first attempt to carry out large-scale measurement-driven analysis on Tumblr.", "Many of the world's most popular websites catalyze their growth through invitations from existing members. New members can then in turn issue invitations, and so on, creating cascades of member signups that can spread on a global scale. Although these diffusive invitation processes are critical to the popularity and growth of many websites, they have rarely been studied, and their properties remain elusive. For instance, it is not known how viral these cascades structures are, how cascades grow over time, or how diffusive growth affects the resulting distribution of member characteristics present on the site. In this paper, we study the diffusion of LinkedIn, an online professional network comprising over 332 million members, a large fraction of whom joined the site as part of a signup cascade. First we analyze the structural patterns of these signup cascades, and find them to be qualitatively different from previously studied information diffusion cascades. We also examine how signup cascades grow over time, and observe that diffusion via invitations on LinkedIn occurs over much longer timescales than are typically associated with other types of online diffusion. Finally, we connect the cascade structures with rich individual-level attribute data to investigate the interplay between the two. Using novel techniques to study the role of homophily in diffusion, we find striking differences between the local, edge-wise homophily and the global, cascade-level homophily we observe in our data, suggesting that signup cascades form surprisingly coherent groups of members.", "We investigated the differential diffusion of all of the verified true and false news stories distributed on Twitter from 2006 to 2017. The data comprise 126,000 stories tweeted by 3 million people more than 4.5 million times. We classified news as true or false using information from six independent fact-checking organizations that exhibited 95 to 98 agreement on the classifications. Falsehood diffused significantly farther, faster, deeper, and more broadly than the truth in all categories of information, and the effects were more pronounced for false political news than for false news about terrorism, natural disasters, science, urban legends, or financial information. We found that false news was more novel than true news, which suggests that people were more likely to share novel information. Whereas false stories inspired fear, disgust, and surprise in replies, true stories inspired anticipation, sadness, joy, and trust. Contrary to conventional wisdom, robots accelerated the spread of true and false news at the same rate, implying that false news spreads more than the truth because humans, not robots, are more likely to spread it.", "", "When a message, such as a piece of news, spreads in social networks, how can we classify it into categories of interests, such as genuine or fake news? Classification of social media content is a fundamental task for social media mining, and most existing methods regard it as a text categorization problem and mainly focus on using content features, such as words and hashtags. However, for many emerging applications like fake news and rumor detection, it is very challenging, if not impossible, to identify useful features from content. For example, intentional spreaders of fake news may manipulate the content to make it look like real news. To address this problem, this paper concentrates on modeling the propagation of messages in a social network. Specifically, we propose a novel approach, TraceMiner, to (1) infer embeddings of social media users with social network structures; and (2) utilize an LSTM-RNN to represent and classify propagation pathways of a message. Since content information is sparse and noisy on social media, adopting TraceMiner allows to provide a high degree of classification accuracy even in the absence of content information. Experimental results on real-world datasets show the superiority over state-of-the-art approaches on the task of fake news detection and news categorization.", "Online social networks enable collectives of users to create and share content at scale. The diffusion of content through the network, and the resulting information cascades, are phenomena that have been widely investigated on various platforms, which facilitate information diffusion using diverse technical mechanisms, user interfaces and incentives. This paper focuses on Tumblr, an online microblogging social network with a core 'reblogging' functionality that allows information to diffuse across its network by appearing on multiple user blogs. The formation of any cascade network is visible as a list of reblogging events attached as notes to each appearance of the post in the cascade. In this paper, we examine cascade networks on Tumblr, recreated from the series of diffusion events, and analyse them from structural and temporal perspectives. To achieve this, we utilise a cascade construction model that create cascade networks, overcoming problems of a lack of contextual information and missing degraded data. Finally, we compare cascades in Tumblr with those appearing on other social network platforms. Our analysis shows that popular content on Tumblr creates 'large' cascades that are deep, branching into a large number of separate and long paths, having a consistent number of reblogs at each depth and at each given time.", "In 2010, a paper entitled \"From Obscurity to Prominence in Minutes: Political Speech and Real-time search\" won the Best Paper Prize of the WebSci'10 conference. Among its findings were the discovery and documentation of what was labeled a \"Twitter bomb\", an organized effort to spread misinformation about the democratic candidate Martha Coakley through anonymous Twitter accounts. In this paper, after summarizing the details of that event, we outline the recipe of how social networks are used to spread misinformation. One of the most important steps in such a recipe is the \"infiltration\" of a community of users who are already engaged in conversations about a topic, to use them as organic spreaders of misinformation in their extended subnetworks. Then, we take this misinformation spreading recipe and indicate how it was successfully used to spread fake news during the 2016 U.S. Presidential Election. The main differences between the scenarios are the use of Facebook instead of Twitter, and the respective motivations (in 2010: political influence; in 2016: financial benefit through online advertising). After situating these events in the broader context of exploiting the Web, we seize this opportunity to address limitations of the reach of research findings and to start a conversation about how communities of researchers can in- crease their impact on real-world societ al issues.", "The wide availability of user-provided content in online social media facilitates the aggregation of people around common interests, worldviews, and narratives. However, the World Wide Web (WWW) also allows for the rapid dissemination of unsubstantiated rumors and conspiracy theories that often elicit rapid, large, but naive social responses such as the recent case of Jade Helm 15––where a simple military exercise turned out to be perceived as the beginning of a new civil war in the United States. In this work, we address the determinants governing misinformation spreading through a thorough quantitative analysis. In particular, we focus on how Facebook users consume information related to two distinct narratives: scientific and conspiracy news. We find that, although consumers of scientific and conspiracy stories present similar consumption patterns with respect to content, cascade dynamics differ. Selective exposure to content is the primary driver of content diffusion and generates the formation of homogeneous clusters, i.e., “echo chambers.” Indeed, homogeneity appears to be the primary driver for the diffusion of contents and each echo chamber has its own cascade dynamics. Finally, we introduce a data-driven percolation model mimicking rumor spreading and we show that homogeneity and polarization are the main determinants for predicting cascades’ size.", "Cascades of information-sharing are a primary mechanism by which content reaches its audience on social media, and an active line of research has studied how such cascades, which form as content is reshared from person to person, develop and subside. In this paper, we perform a large-scale analysis of cascades on Facebook over significantly longer time scales, and find that a more complex picture emerges, in which many large cascades recur, exhibiting multiple bursts of popularity with periods of quiescence in between. We characterize recurrence by measuring the time elapsed between bursts, their overlap and proximity in the social network, and the diversity in the demographics of individuals participating in each peak. We discover that content virality, as revealed by its initial popularity, is a main driver of recurrence, with the availability of multiple copies of that content helping to spark new bursts. Still, beyond a certain popularity of content, the rate of recurrence drops as cascades start exhausting the population of interested individuals. We reproduce these observed patterns in a simple model of content recurrence simulated on a real social network. Using only characteristics of a cascade's initial burst, we demonstrate strong performance in predicting whether it will recur in the future.", "", "Viral products and ideas are intuitively understood to grow through a person-to-person diffusion process analogous to the spread of an infectious disease; however, until recently it has been prohibitively difficult to directly observe purportedly viral events, and thus to rigorously quantify or characterize their structural properties. Here we propose a formal measure of what we label “structural virality” that interpolates between two conceptual extremes: content that gains its popularity through a single, large broadcast and that which grows through multiple generations with any one individual directly responsible for only a fraction of the total adoption. We use this notion of structural virality to analyze a unique data set of a billion diffusion events on Twitter, including the propagation of news stories, videos, images, and petitions. We find that across all domains and all sizes of events, online diffusion is characterized by surprising structural diversity; that is, popular events regularly grow via both broadcast and viral mechanisms, as well as essentially all conceivable combinations of the two. Nevertheless, we find that structural virality is typically low, and remains so independent of size, suggesting that popularity is largely driven by the size of the largest broadcast. Finally, we attempt to replicate these findings with a model of contagion characterized by a low infection rate spreading on a scale-free network. We find that although several of our empirical findings are consistent with such a model, it fails to replicate the observed diversity of structural virality, thereby suggesting new directions for future modeling efforts. This paper was accepted by Lorin Hitt, information systems.", "Many previous techniques identify trending topics in social media, even topics that are not pre-defined. We present a technique to identify trending rumors, which we define as topics that include disputed factual claims. Putting aside any attempt to assess whether the rumors are true or false, it is valuable to identify trending rumors as early as possible. It is extremely difficult to accurately classify whether every individual post is or is not making a disputed factual claim. We are able to identify trending rumors by recasting the problem as finding entire clusters of posts whose topic is a disputed factual claim. The key insight is that when there is a rumor, even though most posts do not raise questions about it, there may be a few that do. If we can find signature text phrases that are used by a few people to express skepticism about factual claims and are rarely used to express anything else, we can use those as detectors for rumor clusters. Indeed, we have found a few phrases that seem to be used exactly that way, including: \"Is this true?\", \"Really?\", and \"What?\". Relatively few posts related to any particular rumor use any of these enquiry phrases, but lots of rumor diffusion processes have some posts that do and have them quite early in the diffusion. We have developed a technique based on searching for the enquiry phrases, clustering similar posts together, and then collecting related posts that do not contain these simple phrases. We then rank the clusters by their likelihood of really containing a disputed factual claim. The detector, which searches for the very rare but very informative phrases, combined with clustering and a classifier on the clusters, yields surprisingly good performance. On a typical day of Twitter, about a third of the top 50 clusters were judged to be rumors, a high enough precision that human analysts might be willing to sift through them.", "On many social networking web sites such as Facebook and Twitter, resharing or reposting functionality allows users to share others' content with their own friends or followers. As content is reshared from user to user, large cascades of reshares can form. While a growing body of research has focused on analyzing and characterizing such cascades, a recent, parallel line of work has argued that the future trajectory of a cascade may be inherently unpredictable. In this work, we develop a framework for addressing cascade prediction problems. On a large sample of photo reshare cascades on Facebook, we find strong performance in predicting whether a cascade will continue to grow in the future. We find that the relative growth of a cascade becomes more predictable as we observe more of its reshares, that temporal and structural features are key predictors of cascade size, and that initially, breadth, rather than depth in a cascade is a better indicator of larger cascades. This prediction performance is robust in the sense that multiple distinct classes of features all achieve similar performance. We also discover that temporal features are predictive of a cascade's eventual shape. Observing independent cascades of the same content, we find that while these cascades differ greatly in size, we are still able to predict which ends up the largest.", "Online social network platforms provide built-in functionalities that allow users to share information through their social connections. The study of information diffusion focusses on analysing how the spread of information enables us to understand the ways that users behave and interact online. Information diffusion manifests itself on social networks in the form of cascades of information, which are structural representations of diffusion events. The characteristics of information diffusion differ between online social networks partly depending on the capabilities of each. Tumblr is an online social network platform; it allows users to reblog more than once, to add a comment with a reblog, and to delete reblogs which presents certain challenges in the study of diffusion on this network. In this paper, we will identify and address those challenges, and we will examine the effects of these functionalities on users' behaviour and consequently on the structural characteristics of cascades on Tumblr.", "Characterizing information diffusion on social platforms like Twitter enables us to understand the properties of underlying media and model communication patterns. As Twitter gains in popularity, it has also become a venue to broadcast rumors and misinformation. We use epidemiological models to characterize information cascades in twitter resulting from both news and rumors. Specifically, we use the SEIZ enhanced epidemic model that explicitly recognizes skeptics to characterize eight events across the world and spanning a range of event types. We demonstrate that our approach is accurate at capturing diffusion in these events. Our approach can be fruitfully combined with other strategies that use content modeling and graph theoretic features to detect (and possibly disrupt) rumors.", "How do blogs cite and influence each other? How do such links evolve? Does the popularity of old blog posts drop exponentially with time? These are some of the questions that we address in this work. Blogs (weblogs) have become an important medium of information because of their timely publication, ease of use, and wide availability. In fact, they often make headlines, by discussing and discovering evidence about political events and facts. Often blogs link to one another, creating a publicly available record of how information and influence spreads through an underlying social network. Aggregating links from several blog posts creates a directed graph which we analyze to discover the patterns of information propagation in blogspace, and thereby understand the underlying social network. Here we report some surprising findings of the blog linking and information propagation structure, after we analyzed one of the largest available datasets, with 45, 000 blogs and ≈ 2.2 million blog-postings. Our analysis also sheds light on how rumors, viruses, and ideas propagate over social and computer networks." ] }
1812.01693	2903015761	The present online social media platform is afflicted with several issues, with hate speech being on the predominant forefront. The prevalence of online hate speech has fueled horrific real-world hate-crime such as the mass-genocide of Rohingya Muslims, communal violence in Colombo and the recent massacre in the Pittsburgh synagogue. Consequently, It is imperative to understand the diffusion of such hateful content in an online setting. We conduct the first study that analyses the flow and dynamics of posts generated by hateful and non-hateful users on Gab (gab.com) over a massive dataset of 341K users and 21M posts. Our observations confirms that hateful content diffuse farther, wider and faster and have a greater outreach than those of non-hateful users. A deeper inspection into the profiles and network of hateful and non-hateful users reveals that the former are more influential, popular and cohesive. Thus, our research explores the interesting facets of diffusion dynamics of hateful users and broadens our understanding of hate speech in the online world.	There is little research done on Gab. @cite_26 performed the first study in which the author collected and analyzed a large dataset of Gab and found that the site is predominantly used for discussion of news, world events, and politics. They also found that Gab contains 2.4 times most hatspeech as compared to Twitter. @cite_32 also found that Gab is very politically oriented and users who abuse the lack of moderation disseminate hate. @cite_13 perform a large scale measurement study of the meme ecosystem by introducing a novel image processing pipeline. Gab has substantially higher number of posts with racist memes. Gab share hateful and racist memes at a higher rate than manstream communities. Similarly, @cite_20 study millions of comments and images from alt-right web communities like 4chan's Politically Incorrect board ( pol ) and the Twitter clone, Gab and quantify the escalation and spread of antisemitism.	{ "cite_N": [ "@cite_26", "@cite_13", "@cite_32", "@cite_20" ], "mid": [ "2786634640", "2951850250", "2963008590", "2890304913" ], "abstract": [ "Over the past few years, a number of new \"fringe\" communities, like 4chan or certain subreddits, have gained traction on the Web at a rapid pace. However, more often than not, little is known about how they evolve or what kind of activities they attract, despite recent research has shown that they influence how false information reaches mainstream communities. This motivates the need to monitor these communities and analyze their impact on the Web's information ecosystem. In August 2016, a new social network called Gab was created as an alternative to Twitter. It positions itself as putting \"people and free speech first\", welcoming users banned or suspended from other social networks. In this paper, we provide, to the best of our knowledge, the first characterization of Gab. We collect and analyze 22M posts produced by 336K users between August 2016 and January 2018, finding that Gab is predominantly used for the dissemination and discussion of news and world events, and that it attracts alt-right users, conspiracy theorists, and other trolls. We also measure the prevalence of hate speech on the platform, finding it to be much higher than Twitter, but lower than 4chan's Politically Incorrect board.", "Internet memes are increasingly used to sway and manipulate public opinion. This prompts the need to study their propagation, evolution, and influence across the Web. In this paper, we detect and measure the propagation of memes across multiple Web communities, using a processing pipeline based on perceptual hashing and clustering techniques, and a dataset of 160M images from 2.6B posts gathered from Twitter, Reddit, 4chan's Politically Incorrect board ( pol ), and Gab, over the course of 13 months. We group the images posted on fringe Web communities ( pol , Gab, and The_Donald subreddit) into clusters, annotate them using meme metadata obtained from Know Your Meme, and also map images from mainstream communities (Twitter and Reddit) to the clusters. Our analysis provides an assessment of the popularity and diversity of memes in the context of each community, showing, e.g., that racist memes are extremely common in fringe Web communities. We also find a substantial number of politics-related memes on both mainstream and fringe Web communities, supporting media reports that memes might be used to enhance or harm politicians. Finally, we use Hawkes processes to model the interplay between Web communities and quantify their reciprocal influence, finding that pol substantially influences the meme ecosystem with the number of memes it produces, while has a higher success rate in pushing them to other communities.", "The moderation of content in many social media systems, such as Twitter and Facebook, motivated the emergence of a new social network system that promotes free speech, named Gab. Soon after that, Gab has been removed from Google Play Store for violating the company's hate speech policy and it has been rejected by Apple for similar reasons. In this paper we characterize Gab, aiming at understanding who are the users who joined it and what kind of content they share in this system. Our findings show that Gab is a very politically oriented system that hosts banned users from other social networks, some of them due to possible cases of hate speech and association with extremism. We provide the first measurement of news dissemination inside a right-leaning echo chamber, investigating a social media where readers are rarely exposed to content that cuts across ideological lines, but rather are fed with content that reinforces their current political or social views.", "A new wave of growing antisemitism, driven by fringe Web communities, is an increasingly worrying presence in the socio-political realm. The ubiquitous and global nature of the Web has provided tools used by these groups to spread their ideology to the rest of the Internet. Although the study of antisemitism and hate is not new, the scale and rate of change of online data has impacted the efficacy of traditional approaches to measure and understand this worrying trend. In this paper, we present a large-scale, quantitative study of online antisemitism. We collect hundreds of million comments and images from alt-right Web communities like 4chan's Politically Incorrect board ( pol ) and the Twitter clone, Gab. Using scientifically grounded methods, we quantify the escalation and spread of antisemitic memes and rhetoric across the Web. We find the frequency of antisemitic content greatly increases (in some cases more than doubling) after major political events such as the 2016 US Presidential Election and the \"Unite the Right\" rally in Charlottesville. Furthermore, this antisemitism appears in tandem with sharp increases in white ethnic nationalist content on the same communities. We extract semantic embeddings from our corpus of posts and demonstrate how automated techniques can discover and categorize the use of antisemitic terminology. We additionally examine the prevalence and spread of the antisemitic \"Happy Merchant\" meme, and in particular how these fringe communities influence its propagation to more mainstream services like Twitter and Reddit. Taken together, our results provide a data-driven, quantitative framework for understanding online antisemitism. Our open and scientifically grounded methods serve as a framework to augment current qualitative efforts by anti-hate groups, providing new insights into the growth and spread of antisemitism online." ] }
1812.01690	2902776281	Deep learning has significant potential for medical imaging. However, since the incident rate of each disease varies widely, the frequency of classes in a medical image dataset is imbalanced, leading to poor accuracy for such infrequent classes. One possible solution is data augmentation of infrequent classes using synthesized images created by Generative Adversarial Networks (GANs), but conventional GANs also require certain amount of images to learn. To overcome this limitation, here we propose General-to-detailed GAN (GDGAN), serially connected two GANs, one for general labels and the other for detailed labels. GDGAN produced diverse medical images, and the network trained with an augmented dataset outperformed other networks using existing methods with respect to Area-Under-Curve (AUC) of Receiver Operating Characteristic (ROC) curve.	Dataset augmentation for non-medical datasets utilizing GANs has been reported in 2017 @cite_9 . Although the datasets used did not have class imbalance, the augmentation generally improved the performance of deep convolutional networks. The synthesis has been done for relatively small medical datasets as well @cite_0 @cite_4 . They aimed to increase the total number of data in the dataset. Xin, Ekta and Paul tackled the imbalance of medical images by using GAN for a future extraction @cite_3 . However, they tested the effect of GAN only by support vector machine.	{ "cite_N": [ "@cite_0", "@cite_9", "@cite_4", "@cite_3" ], "mid": [ "2752237898", "2770173563", "2794022343", "2798261681" ], "abstract": [ "Currently there is strong interest in data-driven approaches to medical image classification. However, medical imaging data is scarce, expensive, and fraught with legal concerns regarding patient privacy. Typical consent forms only allow for patient data to be used in medical journals or education, meaning the majority of medical data is inaccessible for general public research. We propose a novel, two-stage pipeline for generating synthetic medical images from a pair of generative adversarial networks, tested in practice on retinal fundi images. We develop a hierarchical generation process to divide the complex image generation task into two parts: geometry and photorealism. We hope researchers will use our pipeline to bring private medical data into the public domain, sparking growth in imaging tasks that have previously relied on the hand-tuning of models. We have begun this initiative through the development of SynthMed, an online repository for synthetic medical images.", "Effective training of neural networks requires much data. In the low-data regime, parameters are underdetermined, and learnt networks generalise poorly. Data Augmentation krizhevsky2012imagenet alleviates this by using existing data more effectively. However standard data augmentation produces only limited plausible alternative data. Given there is potential to generate a much broader set of augmentations, we design and train a generative model to do data augmentation. The model, based on image conditional Generative Adversarial Networks, takes data from a source domain and learns to take any data item and generalise it to generate other within-class data items. As this generative process does not depend on the classes themselves, it can be applied to novel unseen classes of data. We show that a Data Augmentation Generative Adversarial Network (DAGAN) augments standard vanilla classifiers well. We also show a DAGAN can enhance few-shot learning systems such as Matching Networks. We demonstrate these approaches on Omniglot, on EMNIST having learnt the DAGAN on Omniglot, and VGG-Face data. In our experiments we can see over 13 increase in accuracy in the low-data regime experiments in Omniglot (from 69 to 82 ), EMNIST (73.9 to 76 ) and VGG-Face (4.5 to 12 ); in Matching Networks for Omniglot we observe an increase of 0.5 (from 96.9 to 97.4 ) and an increase of 1.8 in EMNIST (from 59.5 to 61.3 ).", "Abstract Deep learning methods, and in particular convolutional neural networks (CNNs), have led to an enormous breakthrough in a wide range of computer vision tasks, primarily by using large-scale annotated datasets. However, obtaining such datasets in the medical domain remains a challenge. In this paper, we present methods for generating synthetic medical images using recently presented deep learning Generative Adversarial Networks (GANs). Furthermore, we show that generated medical images can be used for synthetic data augmentation, and improve the performance of CNN for medical image classification. Our novel method is demonstrated on a limited dataset of computed tomography (CT) images of 182 liver lesions (53 cysts, 64 metastases and 65 hemangiomas). We first exploit GAN architectures for synthesizing high quality liver lesion ROIs. Then we present a novel scheme for liver lesion classification using CNN. Finally, we train the CNN using classic data augmentation and our synthetic data augmentation and compare performance. In addition, we explore the quality of our synthesized examples using visualization and expert assessment. The classification performance using only classic data augmentation yielded 78.6 sensitivity and 88.4 specificity. By adding the synthetic data augmentation the results increased to 85.7 sensitivity and 92.4 specificity. We believe that this approach to synthetic data augmentation can generalize to other medical classification applications and thus support radiologists’ efforts to improve diagnosis.", "Melanoma is a curable aggressive skin cancer if detected early. Typically, the diagnosis involves initial screening with subsequent biopsy and histopathological examination if necessary. Computer aided diagnosis offers an objective score that is independent of clinical experience and the potential to lower the workload of a dermatologist. In the recent past, success of deep learning algorithms in the field of general computer vision has motivated successful application of supervised deep learning methods in computer aided melanoma recognition. However, large quantities of labeled images are required to make further improvements on the supervised method. A good annotation generally requires clinical and histological confirmation, which requires significant effort. In an attempt to alleviate this constraint, we propose to use categorical generative adversarial network to automatically learn the feature representation of dermoscopy images in an unsupervised and semi-supervised manner. Thorough experiments on ISIC 2016 skin lesion chal- lenge demonstrate that the proposed feature learning method has achieved an average precision score of 0.424 with only 140 labeled images. Moreover, the proposed method is also capable of generating real-world like dermoscopy images." ] }
1812.01894	2786111591	Conventionally, convolutional neural networks (CNNs) process different images with the same set of filters. However, the variations in images pose a challenge to this fashion. In this paper, we propose to generate sample-specific filters for convolutional layers in the forward pass. Since the filters are generated on-the-fly, the model becomes more flexible and can better fit the training data compared to traditional CNNs. In order to obtain sample-specific features, we extract the intermediate feature maps from an autoencoder. As filters are usually high dimensional, we propose to learn a set of coefficients instead of a set of filters. These coefficients are used to linearly combine the base filters from a filter repository to generate the final filters for a CNN. The proposed method is evaluated on MNIST, MTFL and CIFAR10 datasets. Experiment results demonstrate that the classification accuracy of the baseline model can be improved by using the proposed filter generation method.	@cite_14 and @cite_13 also propose similar frameworks for combining decision forests and deep CNNs. Those hybrid models fuse the high representation learning capability of CNNs and the computation efficiency of decision trees.	{ "cite_N": [ "@cite_14", "@cite_13" ], "mid": [ "2290283816", "2622028207" ], "abstract": [ "This paper investigates the connections between two state of the art classifiers: decision forests (DFs, including decision jungles) and convolutional neural networks (CNNs). Decision forests are computationally efficient thanks to their conditional computation property (computation is confined to only a small region of the tree, the nodes along a single branch). CNNs achieve state of the art accuracy, thanks to their representation learning capabilities. We present a systematic analysis of how to fuse conditional computation with representation learning and achieve a continuum of hybrid models with different ratios of accuracy vs. efficiency. We call this new family of hybrid models conditional networks. Conditional networks can be thought of as: i) decision trees augmented with data transformation operators, or ii) CNNs, with block-diagonal sparse weight matrices, and explicit data routing functions. Experimental validation is performed on the common task of image classification on both the CIFAR and Imagenet datasets. Compared to state of the art CNNs, our hybrid models yield the same accuracy with a fraction of the compute cost and much smaller number of parameters.", "We propose a novel method called deep convolutional decision jungle (CDJ) and its learning algorithm for image classification. The CDJ maintains the structure of standard convolutional neural networks (CNNs), i.e. multiple layers of multiple response maps fully connected. Each response map-or node-in both the convolutional and fully-connected layers selectively respond to class labels s.t. each data sample travels via a specific soft route of those activated nodes. The proposed method CDJ automatically learns features, whereas decision forests and jungles require pre-defined feature sets. Compared to CNNs, the method embeds the benefits of using data-dependent discriminative functions, which better handles multi-modal heterogeneous data; further,the method offers more diverse sparse network responses, which in turn can be used for cost-effective learning classification. The network is learnt by combining conventional softmax and proposed entropy losses in each layer. The entropy loss,as used in decision tree growing, measures the purity of data activation according to the class label distribution. The back-propagation rule for the proposed loss function is derived from stochastic gradient descent (SGD) optimization of CNNs. We show that our proposed method outperforms state-of-the-art methods on three public image classification benchmarks and one face verification dataset. We also demonstrate the use of auxiliary data labels, when available, which helps our method to learn more discriminative routing and representations and leads to improved classification." ] }
1812.01825	2902709966	Many reality tasks such as robot coordination can be naturally modelled as multi-agent cooperative system where the rewards are sparse. This paper focuses on learning decentralized policies for such tasks using sub-optimal demonstration. To learn the multi-agent cooperation effectively and tackle the sub-optimality of demonstration, a self-improving learning method is proposed: On the one hand, the centralized state-action values are initialized by the demonstration and updated by the learned decentralized policy to improve the sub-optimality. On the other hand, the Nash Equilibrium are found by the current state-action value and are used as a guide to learn the policy. The proposed method is evaluated on the combat RTS games which requires a high level of multi-agent cooperation. Extensive experimental results on various combat scenarios demonstrate that the proposed method can learn multi-agent cooperation effectively. It significantly outperforms many state-of-the-art demonstration based approaches.	methods have historically been applied in many settings @cite_31 @cite_18 @cite_31 . They have been restricted to tabular methods and simple environments. Motivated from the success of deep RL @cite_14 where value policy is approximated by deep neural networks, the deep MARL methods can scale to high dimensional input and action spaces @cite_33 . Independent Q-learning @cite_19 is proposed to learn decentralised value functions or policies for each agent independently. It is extended to deep neural networks in @cite_33 by combining with DQN @cite_14 . Some methods @cite_6 @cite_17 are proposed to address learning stabilisation following this idea. However, these methods always learn each agent independently and ignore the cooperation among agents.	{ "cite_N": [ "@cite_18", "@cite_14", "@cite_33", "@cite_6", "@cite_19", "@cite_31", "@cite_17" ], "mid": [ "2099618002", "2145339207", "2963658727", "2949201811", "1641379095", "1518858799", "" ], "abstract": [ "Multiagent systems are rapidly finding applications in a variety of domains, including robotics, distributed control, telecommunications, and economics. The complexity of many tasks arising in these domains makes them difficult to solve with preprogrammed agent behaviors. The agents must, instead, discover a solution on their own, using learning. A significant part of the research on multiagent learning concerns reinforcement learning techniques. This paper provides a comprehensive survey of multiagent reinforcement learning (MARL). A central issue in the field is the formal statement of the multiagent learning goal. Different viewpoints on this issue have led to the proposal of many different goals, among which two focal points can be distinguished: stability of the agents' learning dynamics, and adaptation to the changing behavior of the other agents. The MARL algorithms described in the literature aim---either explicitly or implicitly---at one of these two goals or at a combination of both, in a fully cooperative, fully competitive, or more general setting. A representative selection of these algorithms is discussed in detail in this paper, together with the specific issues that arise in each category. Additionally, the benefits and challenges of MARL are described along with some of the problem domains where the MARL techniques have been applied. Finally, an outlook for the field is provided.", "An artificial agent is developed that learns to play a diverse range of classic Atari 2600 computer games directly from sensory experience, achieving a performance comparable to that of an expert human player; this work paves the way to building general-purpose learning algorithms that bridge the divide between perception and action.", "Evolution of cooperation and competition can appear when multiple adaptive agents share a biological, social, or technological niche. In the present work we study how cooperation and competition emerge between autonomous agents that learn by reinforcement while using only their raw visual input as the state representation. In particular, we extend the Deep Q-Learning framework to multiagent environments to investigate the interaction between two learning agents in the well-known video game Pong. By manipulating the classical rewarding scheme of Pong we show how competitive and collaborative behaviors emerge. We also describe the progression from competitive to collaborative behavior when the incentive to cooperate is increased. Finally we show how learning by playing against another adaptive agent, instead of against a hard-wired algorithm, results in more robust strategies. The present work shows that Deep Q-Networks can become a useful tool for studying decentralized learning of multiagent systems coping with high-dimensional environments.", "Many real-world problems, such as network packet routing and urban traffic control, are naturally modeled as multi-agent reinforcement learning (RL) problems. However, existing multi-agent RL methods typically scale poorly in the problem size. Therefore, a key challenge is to translate the success of deep learning on single-agent RL to the multi-agent setting. A major stumbling block is that independent Q-learning, the most popular multi-agent RL method, introduces nonstationarity that makes it incompatible with the experience replay memory on which deep Q-learning relies. This paper proposes two methods that address this problem: 1) using a multi-agent variant of importance sampling to naturally decay obsolete data and 2) conditioning each agent's value function on a fingerprint that disambiguates the age of the data sampled from the replay memory. Results on a challenging decentralised variant of StarCraft unit micromanagement confirm that these methods enable the successful combination of experience replay with multi-agent RL.", "Intelligent human agents exist in a cooperative social environment that facilitates learning. They learn not only by trial-and-error, but also through cooperation by sharing instantaneous information, episodic experience, and learned knowledge. The key investigations of this paper are, “Given the same number of reinforcement learning agents, will cooperative agents outperform independent agents who do not communicate during learning?” and “What is the price for such cooperation?” Using independent agents as a benchmark, cooperative agents are studied in following ways: (1) sharing sensation, (2) sharing episodes, and (3) sharing learned policies. This paper shows that (a) additional sensation from another agent is beneficial if it can be used efficiently, (b) sharing learned policies or episodes among agents speeds up learning at the cost of communication, and (c) for joint tasks, agents engaging in partnership can significantly outperform independent agents although they may learn slowly in the beginning. These tradeoff's are not just limited to multi-agent reinforcement learning.", "Multiagent systems (MAS) are widely accepted as an important method for solving problems of a distributed nature. A key to the success of MAS is efficient and effective multiagent learning (MAL). The past twenty-five years have seen a great interest and tremendous progress in the field of MAL. This article introduces and overviews this field by presenting its fundamentals, sketching its historical development and describing some key algorithms for MAL. Moreover, main challenges that the field is facing today are indentified.", "" ] }
1812.01387	2903462376	In this paper, we present an accurate yet effective solution for 6D pose estimation from an RGB image. The core of our approach is that we first designate a set of surface points on target object model as keypoints and then train a keypoint detector (KPD) to localize them. Finally a PnP algorithm can recover the 6D pose according to the 2D-3D relationship of keypoints. Different from recent state-of-the-art CNN-based approaches that rely on a time-consuming post-processing procedure, our method can achieve competitive accuracy without any refinement after pose prediction. Meanwhile, we obtain a 30 relative improvement in terms of ADD accuracy among methods without using refinement. Moreover, we succeed in handling heavy occlusion by selecting the most confident keypoints to recover the 6D pose. For the sake of reproducibility, we will make our code and models publicly available soon.	Deep learning in 6D pose estimation have demonstrated its great power and potential. @cite_2 points out that one can recover 6D pose from figures by estimating viewpoints or keypoints. The viewpoint based method @cite_13 extends SSD @cite_22 and constructs an end-to-end architecture outputting viewpoints and inplane-rotations. Then it uses the projection relationship to lift the object and further refine estimated poses. @cite_28 creatively learns implicit orientation by Augmented Autoencoder. The keypoints based methods @cite_4 @cite_5 attempt to predict corner points of the bounding boxes and then use PnP algorithm to estimate 6D pose. @cite_12 estimates locations of semantic keypoints instead. Other methods @cite_23 @cite_24 make an effort to train translation and rotation predictors then combine them to get the 6D pose.	{ "cite_N": [ "@cite_4", "@cite_22", "@cite_28", "@cite_24", "@cite_23", "@cite_2", "@cite_5", "@cite_13", "@cite_12" ], "mid": [ "2604236302", "2193145675", "2952188290", "", "2951336016", "2951900634", "", "2488101876", "2949911710" ], "abstract": [ "We introduce a novel method for 3D object detection and pose estimation from color images only. We first use segmentation to detect the objects of interest in 2D even in presence of partial occlusions and cluttered background. By contrast with recent patch-based methods, we rely on a “holistic” approach: We apply to the detected objects a Convolutional Neural Network (CNN) trained to predict their 3D poses in the form of 2D projections of the corners of their 3D bounding boxes. This, however, is not sufficient for handling objects from the recent T-LESS dataset: These objects exhibit an axis of rotational symmetry, and the similarity of two images of such an object under two different poses makes training the CNN challenging. We solve this problem by restricting the range of poses used for training, and by introducing a classifier to identify the range of a pose at run-time before estimating it. We also use an optional additional step that refines the predicted poses. We improve the state-of-the-art on the LINEMOD dataset from 73.7 [2] to 89.3 of correctly registered RGB frames. We are also the first to report results on the Occlusion dataset [1 ] using color images only. We obtain 54 of frames passing the Pose 6D criterion on average on several sequences of the T-LESS dataset, compared to the 67 of the state-of-the-art [10] on the same sequences which uses both color and depth. The full approach is also scalable, as a single network can be trained for multiple objects simultaneously.", "We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. At prediction time, the network generates scores for the presence of each object category in each default box and produces adjustments to the box to better match the object shape. Additionally, the network combines predictions from multiple feature maps with different resolutions to naturally handle objects of various sizes. SSD is simple relative to methods that require object proposals because it completely eliminates proposal generation and subsequent pixel or feature resampling stages and encapsulates all computation in a single network. This makes SSD easy to train and straightforward to integrate into systems that require a detection component. Experimental results on the PASCAL VOC, COCO, and ILSVRC datasets confirm that SSD has competitive accuracy to methods that utilize an additional object proposal step and is much faster, while providing a unified framework for both training and inference. For (300 300 ) input, SSD achieves 74.3 mAP on VOC2007 test at 59 FPS on a Nvidia Titan X and for (512 512 ) input, SSD achieves 76.9 mAP, outperforming a comparable state of the art Faster R-CNN model. Compared to other single stage methods, SSD has much better accuracy even with a smaller input image size. Code is available at https: github.com weiliu89 caffe tree ssd.", "We propose a real-time RGB-based pipeline for object detection and 6D pose estimation. Our novel 3D orientation estimation is based on a variant of the Denoising Autoencoder that is trained on simulated views of a 3D model using Domain Randomization. This so-called Augmented Autoencoder has several advantages over existing methods: It does not require real, pose-annotated training data, generalizes to various test sensors and inherently handles object and view symmetries. Instead of learning an explicit mapping from input images to object poses, it provides an implicit representation of object orientations defined by samples in a latent space. Experiments on the T-LESS and LineMOD datasets show that our method outperforms similar model-based approaches and competes with state-of-the art approaches that require real pose-annotated images.", "", "We present a robust and real-time monocular six degree of freedom relocalization system. Our system trains a convolutional neural network to regress the 6-DOF camera pose from a single RGB image in an end-to-end manner with no need of additional engineering or graph optimisation. The algorithm can operate indoors and outdoors in real time, taking 5ms per frame to compute. It obtains approximately 2m and 6 degree accuracy for large scale outdoor scenes and 0.5m and 10 degree accuracy indoors. This is achieved using an efficient 23 layer deep convnet, demonstrating that convnets can be used to solve complicated out of image plane regression problems. This was made possible by leveraging transfer learning from large scale classification data. We show the convnet localizes from high level features and is robust to difficult lighting, motion blur and different camera intrinsics where point based SIFT registration fails. Furthermore we show how the pose feature that is produced generalizes to other scenes allowing us to regress pose with only a few dozen training examples. PoseNet code, dataset and an online demonstration is available on our project webpage, at this http URL", "We characterize the problem of pose estimation for rigid objects in terms of determining viewpoint to explain coarse pose and keypoint prediction to capture the finer details. We address both these tasks in two different settings - the constrained setting with known bounding boxes and the more challenging detection setting where the aim is to simultaneously detect and correctly estimate pose of objects. We present Convolutional Neural Network based architectures for these and demonstrate that leveraging viewpoint estimates can substantially improve local appearance based keypoint predictions. In addition to achieving significant improvements over state-of-the-art in the above tasks, we analyze the error modes and effect of object characteristics on performance to guide future efforts towards this goal.", "", "We present a 3D object detection method that uses regressed descriptors of locally-sampled RGB-D patches for 6D vote casting. For regression, we employ a convolutional auto-encoder that has been trained on a large collection of random local patches. During testing, scene patch descriptors are matched against a database of synthetic model view patches and cast 6D object votes which are subsequently filtered to refined hypotheses. We evaluate on three datasets to show that our method generalizes well to previously unseen input data, delivers robust detection results that compete with and surpass the state-of-the-art while being scalable in the number of objects.", "This paper presents a novel approach to estimating the continuous six degree of freedom (6-DoF) pose (3D translation and rotation) of an object from a single RGB image. The approach combines semantic keypoints predicted by a convolutional network (convnet) with a deformable shape model. Unlike prior work, we are agnostic to whether the object is textured or textureless, as the convnet learns the optimal representation from the available training image data. Furthermore, the approach can be applied to instance- and class-based pose recovery. Empirically, we show that the proposed approach can accurately recover the 6-DoF object pose for both instance- and class-based scenarios with a cluttered background. For class-based object pose estimation, state-of-the-art accuracy is shown on the large-scale PASCAL3D+ dataset." ] }
1812.01393	2964299816	Scene text detection is an important step in the scene text reading system. The main challenges lie in significantly varied sizes and aspect ratios, arbitrary orientations, and shapes. Driven by the recent progress in deep learning, impressive performances have been achieved for multi-oriented text detection. Yet, the performance drops dramatically in detecting the curved texts due to the limited text representation (e.g., horizontal bounding boxes, rotated rectangles, or quadrilaterals). It is of great interest to detect the curved texts, which are actually very common in natural scenes. In this paper, we present a novel text detector named TextField for detecting irregular scene texts. Specifically, we learn a direction field pointing away from the nearest text boundary to each text point. This direction field is represented by an image of 2D vectors and learned via a fully convolutional neural network. It encodes both binary text mask and direction information used to separate adjacent text instances, which is challenging for the classical segmentation-based approaches. Based on the learned direction field, we apply a simple yet effective morphological-based post-processing to achieve the final detection. The experimental results show that the proposed TextField outperforms the state-of-the-art methods by a large margin (28 and 8 ) on two curved text datasets: Total-Text and SCUT-CTW1500, respectively; TextField also achieves very competitive performance on multi-oriented datasets: ICDAR 2015 and MSRA-TD500. Furthermore, TextField is robust in generalizing unseen datasets.	Scene text detection methods can be roughly classified into specifically engineered and deep learning-based methods. Before the era of deep learning, scene text detector pipelines usually consist of text component extraction and filtering, component grouping, and candidate filtering. The key step is extracting text components based on some engineered features. Maximally Stable Extremal Regions (MSER) @cite_22 and Stroke Width Transform (SWT) @cite_43 are two representative works for text component extraction. Many traditional methods @cite_1 @cite_46 @cite_4 @cite_52 @cite_20 are based on these two algorithms. Other examples of this type are @cite_36 @cite_37 . Most recent methods shift to deep neural networks to extract scene texts. In general, they can be roughly summarized into regression-based and segmentation-based methods. For the regression-based ones, they can be further divided into two categories based on the target to regress: proposal-based and part-based methods.	{ "cite_N": [ "@cite_37", "@cite_4", "@cite_22", "@cite_36", "@cite_1", "@cite_52", "@cite_43", "@cite_46", "@cite_20" ], "mid": [ "1999284580", "2128854450", "2124404372", "1966693245", "1521064364", "117491841", "2142159465", "2061802763", "" ], "abstract": [ "Text in an image provides vital information for interpreting its contents, and text in a scene can aid a variety of tasks from navigation to obstacle avoidance and odometry. Despite its value, however, detecting general text in images remains a challenging research problem. Motivated by the need to consider the widely varying forms of natural text, we propose a bottom-up approach to the problem, which reflects the characterness of an image region. In this sense, our approach mirrors the move from saliency detection methods to measures of objectness. In order to measure the characterness, we develop three novel cues that are tailored for character detection and a Bayesian method for their integration. Because text is made up of sets of characters, we then design a Markov random field model so as to exploit the inherent dependencies between characters. We experimentally demonstrate the effectiveness of our characterness cues as well as the advantage of Bayesian multicue integration. The proposed text detector outperforms state-of-the-art methods on a few benchmark scene text detection data sets. We also show that our measurement of characterness is superior than state-of-the-art saliency detection models when applied to the same task.", "In this paper, we present a new approach for text localization in natural images, by discriminating text and non-text regions at three levels: pixel, component and text line levels. Firstly, a powerful low-level filter called the Stroke Feature Transform (SFT) is proposed, which extends the widely-used Stroke Width Transform (SWT) by incorporating color cues of text pixels, leading to significantly enhanced performance on inter-component separation and intra-component connection. Secondly, based on the output of SFT, we apply two classifiers, a text component classifier and a text-line classifier, sequentially to extract text regions, eliminating the heuristic procedures that are commonly used in previous approaches. The two classifiers are built upon two novel Text Covariance Descriptors (TCDs) that encode both the heuristic properties and the statistical characteristics of text stokes. Finally, text regions are located by simply thresholding the text-line confident map. Our method was evaluated on two benchmark datasets: ICDAR 2005 and ICDAR 2011, and the corresponding F-measure values are 0.72 and 0.73, respectively, surpassing previous methods in accuracy by a large margin.", "Abstract The wide-baseline stereo problem, i.e. the problem of establishing correspondences between a pair of images taken from different viewpoints is studied. A new set of image elements that are put into correspondence, the so called extremal regions , is introduced. Extremal regions possess highly desirable properties: the set is closed under (1) continuous (and thus projective) transformation of image coordinates and (2) monotonic transformation of image intensities. An efficient (near linear complexity) and practically fast detection algorithm (near frame rate) is presented for an affinely invariant stable subset of extremal regions, the maximally stable extremal regions (MSER). A new robust similarity measure for establishing tentative correspondences is proposed. The robustness ensures that invariants from multiple measurement regions (regions obtained by invariant constructions from extremal regions), some that are significantly larger (and hence discriminative) than the MSERs, may be used to establish tentative correspondences. The high utility of MSERs, multiple measurement regions and the robust metric is demonstrated in wide-baseline experiments on image pairs from both indoor and outdoor scenes. Significant change of scale (3.5×), illumination conditions, out-of-plane rotation, occlusion, locally anisotropic scale change and 3D translation of the viewpoint are all present in the test problems. Good estimates of epipolar geometry (average distance from corresponding points to the epipolar line below 0.09 of the inter-pixel distance) are obtained.", "Scene text extraction methodologies are usually based in classification of individual regions or patches, using a priori knowledge for a given script or language. Human perception of text, on the other hand, is based on perceptual organisation through which text emerges as a perceptually significant group of atomic objects. Therefore humans are able to detect text even in languages and scripts never seen before. In this paper, we argue that the text extraction problem could be posed as the detection of meaningful groups of regions. We present a method built around a perceptual organisation framework that exploits collaboration of proximity and similarity laws to create text-group hypotheses. Experiments demonstrate that our algorithm is competitive with state of the art approaches on a standard dataset covering text in variable orientations and two languages.", "We present a method for spotting words in the wild, i.e., in real images taken in unconstrained environments. Text found in the wild has a surprising range of difficulty. At one end of the spectrum, Optical Character Recognition (OCR) applied to scanned pages of well formatted printed text is one of the most successful applications of computer vision to date. At the other extreme lie visual CAPTCHAs - text that is constructed explicitly to fool computer vision algorithms. Both tasks involve recognizing text, yet one is nearly solved while the other remains extremely challenging. In this work, we argue that the appearance of words in the wild spans this range of difficulties and propose a new word recognition approach based on state-of-the-art methods from generic object recognition, in which we consider object categories to be the words themselves. We compare performance of leading OCR engines - one open source and one proprietary - with our new approach on the ICDAR Robust Reading data set and a new word spotting data set we introduce in this paper: the Street View Text data set. We show improvements of up to 16 on the data sets, demonstrating the feasibility of a new approach to a seemingly old problem.", "Maximally Stable Extremal Regions (MSERs) have achieved great success in scene text detection. However, this low-level pixel operation inherently limits its capability for handling complex text information efficiently (e. g. connections between text or background components), leading to the difficulty in distinguishing texts from background components. In this paper, we propose a novel framework to tackle this problem by leveraging the high capability of convolutional neural network (CNN). In contrast to recent methods using a set of low-level heuristic features, the CNN network is capable of learning high-level features to robustly identify text components from text-like outliers (e.g. bikes, windows, or leaves). Our approach takes advantages of both MSERs and sliding-window based methods. The MSERs operator dramatically reduces the number of windows scanned and enhances detection of the low-quality texts. While the sliding-window with CNN is applied to correctly separate the connections of multiple characters in components. The proposed system achieved strong robustness against a number of extreme text variations and serious real-world problems. It was evaluated on the ICDAR 2011 benchmark dataset, and achieved over 78 in F-measure, which is significantly higher than previous methods.", "We present a novel image operator that seeks to find the value of stroke width for each image pixel, and demonstrate its use on the task of text detection in natural images. The suggested operator is local and data dependent, which makes it fast and robust enough to eliminate the need for multi-scale computation or scanning windows. Extensive testing shows that the suggested scheme outperforms the latest published algorithms. Its simplicity allows the algorithm to detect texts in many fonts and languages.", "An end-to-end real-time scene text localization and recognition method is presented. The real-time performance is achieved by posing the character detection problem as an efficient sequential selection from the set of Extremal Regions (ERs). The ER detector is robust to blur, illumination, color and texture variation and handles low-contrast text. In the first classification stage, the probability of each ER being a character is estimated using novel features calculated with O(1) complexity per region tested. Only ERs with locally maximal probability are selected for the second stage, where the classification is improved using more computationally expensive features. A highly efficient exhaustive search with feedback loops is then applied to group ERs into words and to select the most probable character segmentation. Finally, text is recognized in an OCR stage trained using synthetic fonts. The method was evaluated on two public datasets. On the ICDAR 2011 dataset, the method achieves state-of-the-art text localization results amongst published methods and it is the first one to report results for end-to-end text recognition. On the more challenging Street View Text dataset, the method achieves state-of-the-art recall. The robustness of the proposed method against noise and low contrast of characters is demonstrated by “false positives” caused by detected watermark text in the dataset.", "" ] }
1812.01393	2964299816	Scene text detection is an important step in the scene text reading system. The main challenges lie in significantly varied sizes and aspect ratios, arbitrary orientations, and shapes. Driven by the recent progress in deep learning, impressive performances have been achieved for multi-oriented text detection. Yet, the performance drops dramatically in detecting the curved texts due to the limited text representation (e.g., horizontal bounding boxes, rotated rectangles, or quadrilaterals). It is of great interest to detect the curved texts, which are actually very common in natural scenes. In this paper, we present a novel text detector named TextField for detecting irregular scene texts. Specifically, we learn a direction field pointing away from the nearest text boundary to each text point. This direction field is represented by an image of 2D vectors and learned via a fully convolutional neural network. It encodes both binary text mask and direction information used to separate adjacent text instances, which is challenging for the classical segmentation-based approaches. Based on the learned direction field, we apply a simple yet effective morphological-based post-processing to achieve the final detection. The experimental results show that the proposed TextField outperforms the state-of-the-art methods by a large margin (28 and 8 ) on two curved text datasets: Total-Text and SCUT-CTW1500, respectively; TextField also achieves very competitive performance on multi-oriented datasets: ICDAR 2015 and MSRA-TD500. Furthermore, TextField is robust in generalizing unseen datasets.	Proposal-based methods are mainly inspired by recent object detection pipelines. TextBoxes @cite_47 directly adapts SSD @cite_42 for scene text detection by using long default boxes and convlutioinal filters to cope with the significantly varied aspect ratios. TextBoxes++ @cite_49 extends TextBoxes by regressing quadrilaterals instead of horizontal bounding boxes. Ma attempt to solve the multi-oriented text detection by adopting Rotated Regional Proposal Network (RRPN) in the pipeline of faster r-cnn. Quadrilateral sliding windows are adopted in @cite_26 to detect multi-oriented texts. Wordsup @cite_60 explores word annotations for character-based text detection. SSTD @cite_41 introduces the attention mechanism by FCN to suppress background interference, improving accurate detection of small texts. Liao propose to apply rotation-invariant and sensitive features for text non-text box classification and regression, respectively, boosting long oriented text detection. @cite_35 proposed instance transformation network by considering geometry-aware information for scene text detection. East @cite_5 and Deep regression @cite_9 both perform per-pixel rotated rectangle or quadrilateral estimation.	{ "cite_N": [ "@cite_35", "@cite_26", "@cite_60", "@cite_41", "@cite_9", "@cite_42", "@cite_49", "@cite_5", "@cite_47" ], "mid": [ "2798450692", "", "2962804639", "2963977642", "2604735854", "2193145675", "2784050770", "2605982830", "2962773189" ], "abstract": [ "Localizing text in the wild is challenging in the situations of complicated geometric layout of the targets like random orientation and large aspect ratio. In this paper, we propose a geometry-aware modeling approach tailored for scene text representation with an end-to-end learning scheme. In our approach, a novel Instance Transformation Network (ITN) is presented to learn the geometry-aware representation encoding the unique geometric configurations of scene text instances with in-network transformation embedding, resulting in a robust and elegant framework to detect words or text lines at one pass. An end-to-end multi-task learning strategy with transformation regression, text non-text classification and coordinates regression is adopted in the ITN. Experiments on the benchmark datasets demonstrate the effectiveness of the proposed approach in detecting scene text in various geometric configurations.", "", "Imagery texts are usually organized as a hierarchy of several visual elements, i.e. characters, words, text lines and text blocks. Among these elements, character is the most basic one for various languages such as Western, Chinese, Japanese, mathematical expression and etc. It is natural and convenient to construct a common text detection engine based on character detectors. However, training character detectors requires a vast of location annotated characters, which are expensive to obtain. Actually, the existing real text datasets are mostly annotated in word or line level. To remedy this dilemma, we propose a weakly supervised framework that can utilize word annotations, either in tight quadrangles or the more loose bounding boxes, for character detector training. When applied in scene text detection, we are thus able to train a robust character detector by exploiting word annotations in the rich large-scale real scene text datasets, e.g. ICDAR15 [19] and COCO-text [39]. The character detector acts as a key role in the pipeline of our text detection engine. It achieves the state-of-the-art performance on several challenging scene text detection benchmarks. We also demonstrate the flexibility of our pipeline by various scenarios, including deformed text detection and math expression recognition.", "We present a novel single-shot text detector that directly outputs word-level bounding boxes in a natural image. We propose an attention mechanism which roughly identifies text regions via an automatically learned attentional map. This substantially suppresses background interference in the convolutional features, which is the key to producing accurate inference of words, particularly at extremely small sizes. This results in a single model that essentially works in a coarse-to-fine manner. It departs from recent FCN-based text detectors which cascade multiple FCN models to achieve an accurate prediction. Furthermore, we develop a hierarchical inception module which efficiently aggregates multi-scale inception features. This enhances local details, and also encodes strong context information, allowing the detector to work reliably on multi-scale and multi-orientation text with single-scale images. Our text detector achieves an F-measure of 77 on the ICDAR 2015 benchmark, advancing the state-of-the-art results in [18, 28]. Demo is available at: http: sstd.whuang.org .", "In this paper, we first provide a new perspective to divide existing high performance object detection methods into direct and indirect regressions. Direct regression performs boundary regression by predicting the offsets from a given point, while indirect regression predicts the offsets from some bounding box proposals. In the context of multioriented scene text detection, we analyze the drawbacks of indirect regression, which covers the state-of-the-art detection structures Faster-RCNN and SSD as instances, and point out the potential superiority of direct regression. To verify this point of view, we propose a deep direct regression based method for multi-oriented scene text detection. Our detection framework is simple and effective with a fully convolutional network and one-step post processing. The fully convolutional network is optimized in an end-to-end way and has bi-task outputs where one is pixel-wise classification between text and non-text, and the other is direct regression to determine the vertex coordinates of quadrilateral text boundaries. The proposed method is particularly beneficial to localize incidental scene texts. On the ICDAR2015 Incidental Scene Text benchmark, our method achieves the F-measure of 81 , which is a new state-ofthe-art and significantly outperforms previous approaches. On other standard datasets with focused scene texts, our method also reaches the state-of-the-art performance.", "We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. At prediction time, the network generates scores for the presence of each object category in each default box and produces adjustments to the box to better match the object shape. Additionally, the network combines predictions from multiple feature maps with different resolutions to naturally handle objects of various sizes. SSD is simple relative to methods that require object proposals because it completely eliminates proposal generation and subsequent pixel or feature resampling stages and encapsulates all computation in a single network. This makes SSD easy to train and straightforward to integrate into systems that require a detection component. Experimental results on the PASCAL VOC, COCO, and ILSVRC datasets confirm that SSD has competitive accuracy to methods that utilize an additional object proposal step and is much faster, while providing a unified framework for both training and inference. For (300 300 ) input, SSD achieves 74.3 mAP on VOC2007 test at 59 FPS on a Nvidia Titan X and for (512 512 ) input, SSD achieves 76.9 mAP, outperforming a comparable state of the art Faster R-CNN model. Compared to other single stage methods, SSD has much better accuracy even with a smaller input image size. Code is available at https: github.com weiliu89 caffe tree ssd.", "Scene text detection is an important step of scene text recognition system and also a challenging problem. Different from general object detections, the main challenges of scene text detection lie on arbitrary orientations, small sizes, and significantly variant aspect ratios of text in natural images. In this paper, we present an end-to-end trainable fast scene text detector, named TextBoxes++, which detects arbitrary-oriented scene text with both high accuracy and efficiency in a single network forward pass. No post-processing other than efficient non-maximum suppression is involved. We have evaluated the proposed TextBoxes++ on four public data sets. In all experiments, TextBoxes++ outperforms competing methods in terms of text localization accuracy and runtime. More specifically, TextBoxes++ achieves an f-measure of 0.817 at 11.6 frames s for 1024 × 1024 ICDAR 2015 incidental text images and an f-measure of 0.5591 at 19.8 frames s for 768 × 768 COCO-Text images. Furthermore, combined with a text recognizer, TextBoxes++ significantly outperforms the state-of-the-art approaches for word spotting and end-to-end text recognition tasks on popular benchmarks. Code is available at: https: github.com MhLiao TextBoxes_plusplus.", "Previous approaches for scene text detection have already achieved promising performances across various benchmarks. However, they usually fall short when dealing with challenging scenarios, even when equipped with deep neural network models, because the overall performance is determined by the interplay of multiple stages and components in the pipelines. In this work, we propose a simple yet powerful pipeline that yields fast and accurate text detection in natural scenes. The pipeline directly predicts words or text lines of arbitrary orientations and quadrilateral shapes in full images, eliminating unnecessary intermediate steps (e.g., candidate aggregation and word partitioning), with a single neural network. The simplicity of our pipeline allows concentrating efforts on designing loss functions and neural network architecture. Experiments on standard datasets including ICDAR 2015, COCO-Text and MSRA-TD500 demonstrate that the proposed algorithm significantly outperforms state-of-the-art methods in terms of both accuracy and efficiency. On the ICDAR 2015 dataset, the proposed algorithm achieves an F-score of 0.7820 at 13.2fps at 720p resolution.", "This paper presents an end-to-end trainable fast scene text detector, named TextBoxes, which detects scene text with both high accuracy and efficiency in a single network forward pass, involving no post-process except for a standard non-maximum suppression. TextBoxes outperforms competing methods in terms of text localization accuracy and is much faster, taking only 0.09s per image in a fast implementation. Furthermore, combined with a text recognizer, TextBoxes significantly outperforms state-of-the-art approaches on word spotting and end-to-end text recognition tasks." ] }
1812.01393	2964299816	Scene text detection is an important step in the scene text reading system. The main challenges lie in significantly varied sizes and aspect ratios, arbitrary orientations, and shapes. Driven by the recent progress in deep learning, impressive performances have been achieved for multi-oriented text detection. Yet, the performance drops dramatically in detecting the curved texts due to the limited text representation (e.g., horizontal bounding boxes, rotated rectangles, or quadrilaterals). It is of great interest to detect the curved texts, which are actually very common in natural scenes. In this paper, we present a novel text detector named TextField for detecting irregular scene texts. Specifically, we learn a direction field pointing away from the nearest text boundary to each text point. This direction field is represented by an image of 2D vectors and learned via a fully convolutional neural network. It encodes both binary text mask and direction information used to separate adjacent text instances, which is challenging for the classical segmentation-based approaches. Based on the learned direction field, we apply a simple yet effective morphological-based post-processing to achieve the final detection. The experimental results show that the proposed TextField outperforms the state-of-the-art methods by a large margin (28 and 8 ) on two curved text datasets: Total-Text and SCUT-CTW1500, respectively; TextField also achieves very competitive performance on multi-oriented datasets: ICDAR 2015 and MSRA-TD500. Furthermore, TextField is robust in generalizing unseen datasets.	Some other regression-based methods tend to regress text parts while predicting the linking relationship between them. @cite_30 , the authors propose a Connectionist Text Proposal Network (CTPN) by first predicting vertical text parts, then adopting a recurrent neural network to link text parts. Shi present a network named SegLink @cite_40 to first detect text parts named text segments while predicting the linking relationship between neighboring text segments. A novel framework named Markov Clustering Network (MCN) was proposed in @cite_21 . In this work, the authors propose to regard an image as a stochastic flow graph, where the flows are strong between text nodes ( , text pixels) but weak for the others. Then a Markov clustering process is applied to form text instances from the predicted flow graph. @cite_58 , propose to first regress four corners of text boxes, followed by a combination of corners and Non-Maximum Suppression (NMS) process to achieve accurate multi-oriented text localization.	{ "cite_N": [ "@cite_30", "@cite_40", "@cite_21", "@cite_58" ], "mid": [ "2519818067", "2605076167", "2963161243", "2963840241" ], "abstract": [ "We propose a novel Connectionist Text Proposal Network (CTPN) that accurately localizes text lines in natural image. The CTPN detects a text line in a sequence of fine-scale text proposals directly in convolutional feature maps. We develop a vertical anchor mechanism that jointly predicts location and text non-text score of each fixed-width proposal, considerably improving localization accuracy. The sequential proposals are naturally connected by a recurrent neural network, which is seamlessly incorporated into the convolutional network, resulting in an end-to-end trainable model. This allows the CTPN to explore rich context information of image, making it powerful to detect extremely ambiguous text. The CTPN works reliably on multi-scale and multi-language text without further post-processing, departing from previous bottom-up methods requiring multi-step post filtering. It achieves 0.88 and 0.61 F-measure on the ICDAR 2013 and 2015 benchmarks, surpassing recent results [8, 35] by a large margin. The CTPN is computationally efficient with 0.14 s image, by using the very deep VGG16 model [27]. Online demo is available: http: textdet.com .", "Most state-of-the-art text detection methods are specific to horizontal Latin text and are not fast enough for real-time applications. We introduce Segment Linking (SegLink), an oriented text detection method. The main idea is to decompose text into two locally detectable elements, namely segments and links. A segment is an oriented box covering a part of a word or text line, A link connects two adjacent segments, indicating that they belong to the same word or text line. Both elements are detected densely at multiple scales by an end-to-end trained, fully-convolutional neural network. Final detections are produced by combining segments connected by links. Compared with previous methods, SegLink improves along the dimensions of accuracy, speed, and ease of training. It achieves an f-measure of 75.0 on the standard ICDAR 2015 Incidental (Challenge 4) benchmark, outperforming the previous best by a large margin. It runs at over 20 FPS on 512x512 images. Moreover, without modification, SegLink is able to detect long lines of non-Latin text, such as Chinese.", "A novel framework named Markov Clustering Network (MCN) is proposed for fast and robust scene text detection. MCN predicts instance-level bounding boxes by firstly converting an image into a Stochastic Flow Graph (SFG) and then performing Markov Clustering on this graph. Our method can detect text objects with arbitrary size and orientation without prior knowledge of object size. The stochastic flow graph encode objects' local correlation and semantic information. An object is modeled as strongly connected nodes, which allows flexible bottom-up detection for scale-varying and rotated objects. MCN generates bounding boxes without using Non-Maximum Suppression, and it can be fully parallelized on GPUs. The evaluation on public benchmarks shows that our method outperforms the existing methods by a large margin in detecting multi-oriented text objects. MCN achieves new state-of-art performance on challenging MSRA-TD500 dataset with precision of 0.88, recall of 0.79 and F-score of 0.83. Also, MCN achieves realtime inference with frame rate of 34 FPS, which is 1.5 A— speedup when compared with the fastest scene text detection algorithm.", "Previous deep learning based state-of-the-art scene text detection methods can be roughly classified into two categories. The first category treats scene text as a type of general objects and follows general object detection paradigm to localize scene text by regressing the text box locations, but troubled by the arbitrary-orientation and large aspect ratios of scene text. The second one segments text regions directly, but mostly needs complex post processing. In this paper, we present a method that combines the ideas of the two types of methods while avoiding their shortcomings. We propose to detect scene text by localizing corner points of text bounding boxes and segmenting text regions in relative positions. In inference stage, candidate boxes are generated by sampling and grouping corner points, which are further scored by segmentation maps and suppressed by NMS. Compared with previous methods, our method can handle long oriented text naturally and doesn't need complex post processing. The experiments on ICDAR2013, ICDAR2015, MSRA-TD500, MLT and COCO-Text demonstrate that the proposed algorithm achieves better or comparable results in both accuracy and efficiency. Based on VGG16, it achieves an F-measure of 84.3 on ICDAR2015 and 81.5 on MSRA-TD500." ] }
1812.01393	2964299816	Scene text detection is an important step in the scene text reading system. The main challenges lie in significantly varied sizes and aspect ratios, arbitrary orientations, and shapes. Driven by the recent progress in deep learning, impressive performances have been achieved for multi-oriented text detection. Yet, the performance drops dramatically in detecting the curved texts due to the limited text representation (e.g., horizontal bounding boxes, rotated rectangles, or quadrilaterals). It is of great interest to detect the curved texts, which are actually very common in natural scenes. In this paper, we present a novel text detector named TextField for detecting irregular scene texts. Specifically, we learn a direction field pointing away from the nearest text boundary to each text point. This direction field is represented by an image of 2D vectors and learned via a fully convolutional neural network. It encodes both binary text mask and direction information used to separate adjacent text instances, which is challenging for the classical segmentation-based approaches. Based on the learned direction field, we apply a simple yet effective morphological-based post-processing to achieve the final detection. The experimental results show that the proposed TextField outperforms the state-of-the-art methods by a large margin (28 and 8 ) on two curved text datasets: Total-Text and SCUT-CTW1500, respectively; TextField also achieves very competitive performance on multi-oriented datasets: ICDAR 2015 and MSRA-TD500. Furthermore, TextField is robust in generalizing unseen datasets.	Segmentation-based approaches regard text detection as a text area segmentation problem, which is usually achieved via Fully Convolutional Neural Network (FCN). They mainly differ in how to post-process the predicted text regions into words or text lines. @cite_50 , Zhang adopted an FCN to estimate text blocks, on which candidate characters are extracted using MSER. Then they use traditional grouping and filtering strategies to achieve multi-oriented text detection. In addition to text block (word or line) prediction, Yao @cite_59 also propose to predict both individual characters and the orientation of text boxes via an FCN in a holistic fashion. Then a grouping process based on the three estimated properties of text yields the text detection. Ch'ng fine-tunes DeconvNet @cite_13 to achieve curved text detection. @cite_24 , the authors consider the text detection as an instance segmentation problem using multi-scale image inputs. They adopted an FCN to predict text blocks, followed by two CNN branches predicting text lines and instance-aware segmentations from the estimated text blocks. Wu @cite_48 introduce text border in addition to text non-text segmentation, which results in a three-class semantic segmentation, facilitating the separation of adjacent text instances.	{ "cite_N": [ "@cite_48", "@cite_24", "@cite_59", "@cite_50", "@cite_13" ], "mid": [ "2776766448", "2740819638", "2464918637", "", "1745334888" ], "abstract": [ "In this paper we propose a new solution to the text detection problem via border learning. Specifically, we make four major contributions: 1) We analyze the insufficiencies of the classic non-text and text settings for text detection. 2) We introduce the border class to the text detection problem for the first time, and validate that the decoding process is largely simplified with the help of text border. 3) We collect and release a new text detection PPT dataset containing 10,692 images with non-text, border, and text annotations. 4) We develop a lightweight (only 0.28M parameters), fully convolutional network (FCN) to effectively learn borders in text images. The results of our extensive experiments show that the proposed solution achieves comparable performance, and often outperforms state-of-theart approaches on standard benchmarks–even though our solution only requires minimal post-processing to parse a bounding box from a detected text map, while others often require heavy post-processing.", "Scene text detection has attracted great attention these years. Text potentially exist in a wide variety of images or videos and play an important role in understanding the scene. In this paper, we present a novel text detection algorithm which is composed of two cascaded steps: (1) a multi-scale fully convolutional neural network (FCN) is proposed to extract text block regions, (2) a novel instance (word or line) aware segmentation is designed to further remove false positives and obtain word instances. The proposed algorithm can accurately localize word or text line in arbitrary orientations, including curved text lines which cannot be handled in a lot of other frameworks. Our algorithm achieved state-of-the-art performance in ICDAR 2013 (IC13), ICDAR 2015 (IC15) and CUTE80 and Street View Text (SVT) benchmark datasets.", "Recently, scene text detection has become an active research topic in computer vision and document analysis, because of its great importance and significant challenge. However, vast majority of the existing methods detect text within local regions, typically through extracting character, word or line level candidates followed by candidate aggregation and false positive elimination, which potentially exclude the effect of wide-scope and long-range contextual cues in the scene. To take full advantage of the rich information available in the whole natural image, we propose to localize text in a holistic manner, by casting scene text detection as a semantic segmentation problem. The proposed algorithm directly runs on full images and produces global, pixel-wise prediction maps, in which detections are subsequently formed. To better make use of the properties of text, three types of information regarding text region, individual characters and their relationship are estimated, with a single Fully Convolutional Network (FCN) model. With such predictions of text properties, the proposed algorithm can simultaneously handle horizontal, multi-oriented and curved text in real-world natural images. The experiments on standard benchmarks, including ICDAR 2013, ICDAR 2015 and MSRA-TD500, demonstrate that the proposed algorithm substantially outperforms previous state-of-the-art approaches. Moreover, we report the first baseline result on the recently-released, large-scale dataset COCO-Text.", "", "We propose a novel semantic segmentation algorithm by learning a deep deconvolution network. We learn the network on top of the convolutional layers adopted from VGG 16-layer net. The deconvolution network is composed of deconvolution and unpooling layers, which identify pixelwise class labels and predict segmentation masks. We apply the trained network to each proposal in an input image, and construct the final semantic segmentation map by combining the results from all proposals in a simple manner. The proposed algorithm mitigates the limitations of the existing methods based on fully convolutional networks by integrating deep deconvolution network and proposal-wise prediction, our segmentation method typically identifies detailed structures and handles objects in multiple scales naturally. Our network demonstrates outstanding performance in PASCAL VOC 2012 dataset, and we achieve the best accuracy (72.5 ) among the methods trained without using Microsoft COCO dataset through ensemble with the fully convolutional network." ] }
1812.01483	2945438069	We introduce Compositional Imitation Learning and Execution (CompILE): a framework for learning reusable, variable-length segments of hierarchically-structured behavior from demonstration data. CompILE uses a novel unsupervised, fully-differentiable sequence segmentation module to learn latent encodings of sequential data that can be re-composed and executed to perform new tasks. Once trained, our model generalizes to sequences of longer length and from environment instances not seen during training. We evaluate CompILE in a challenging 2D multi-task environment and a continuous control task, and show that it can find correct task boundaries and event encodings in an unsupervised manner. Latent codes and associated behavior policies discovered by CompILE can be used by a hierarchical agent, where the high-level policy selects actions in the latent code space, and the low-level, task-specific policies are simply the learned decoders. We found that our CompILE-based agent could learn given only sparse rewards, where agents without task-specific policies struggle.	Our framework is closely related to option discovery @cite_22 @cite_3 @cite_20 @cite_2 , with the main difference being that our inference algorithm is agnostic to what type of option (sub-task) encoding is used. Our framework allows for inference of continuous, discrete or mixed continuous-discrete latent variables within the default VAE @cite_5 @cite_21 setup using the reparameterization trick @cite_5 for low-variance gradient estimation. @cite_20 introduce an EM-based inference algorithm for option discovery in settings similar to ours. Their model, however, has to make several limiting assumptions to be able to use EM for efficient inference: their model is restricted to discrete latent variables and to inference networks that are independent of the position of task boundaries: in their case without recurrency and only dependent on the current state action pair. Option discovery has also been addressed in the context of inverse reinforcement learning (IRL) using generative adversarial networks (GANs) @cite_9 to find structured policies that are close to demonstration sequences @cite_2 . This approach requires being able to interact with the environment for imitation learning, whereas our model is based on BC and works on offline demonstration data.	{ "cite_N": [ "@cite_22", "@cite_9", "@cite_21", "@cite_3", "@cite_2", "@cite_5", "@cite_20" ], "mid": [ "2294748366", "2099471712", "2962897886", "1539716820", "2963221965", "2964121744", "2605016475" ], "abstract": [ "Much recent work in robot learning from demonstration has focused on automatically segmenting continuous task demonstrations into simpler, reusable primitives. However, strong assumptions are often made about how these primitives can be sequenced, limiting the potential for data reuse. We introduce a novel method for discovering semantically grounded primitives and incrementally building and improving a finite-state representation of a task in which various contingencies can arise. Specifically, a Beta Process Autoregressive Hidden Markov Model is used to automatically segment demonstrations into motion categories, which are then further subdivided into semantically grounded states in a finite-state automaton. During replay of the task, a data-driven approach is used to collect additional data where they are most needed through interactive corrections, which are then used to improve the finite-state automaton. Together, this allows for intelligent sequencing of primitives to create novel, adaptive behavior that can be incrementally improved as needed. We demonstrate the utility of this technique on a furniture assembly task using the PR2 mobile manipulator.", "We propose a new framework for estimating generative models via an adversarial process, in which we simultaneously train two models: a generative model G that captures the data distribution, and a discriminative model D that estimates the probability that a sample came from the training data rather than G. The training procedure for G is to maximize the probability of D making a mistake. This framework corresponds to a minimax two-player game. In the space of arbitrary functions G and D, a unique solution exists, with G recovering the training data distribution and D equal to ½ everywhere. In the case where G and D are defined by multilayer perceptrons, the entire system can be trained with backpropagation. There is no need for any Markov chains or unrolled approximate inference networks during either training or generation of samples. Experiments demonstrate the potential of the framework through qualitative and quantitative evaluation of the generated samples.", "We marry ideas from deep neural networks and approximate Bayesian inference to derive a generalised class of deep, directed generative models, endowed with a new algorithm for scalable inference and learning. Our algorithm introduces a recognition model to represent an approximate posterior distribution and uses this for optimisation of a variational lower bound. We develop stochastic backpropagation - rules for gradient backpropagation through stochastic variables - and derive an algorithm that allows for joint optimisation of the parameters of both the generative and recognition models. We demonstrate on several real-world data sets that by using stochastic backpropagation and variational inference, we obtain models that are able to generate realistic samples of data, allow for accurate imputations of missing data, and provide a useful tool for high-dimensional data visualisation.", "Most manipulation tasks can be decomposed into a sequence of phases, where the robot's actions have different effects in each phase. The robot can perform actions to transition between phases and, thus, alter the effects of its actions, e.g. grasp an object in order to then lift it. The robot can thus reach a phase that affords the desired manipulation. In this paper, we present an approach for exploiting the phase structure of tasks in order to learn manipulation skills more efficiently. Starting with human demonstrations, the robot learns a probabilistic model of the phases and the phase transitions. The robot then employs model-based reinforcement learning to create a library of motor primitives for transitioning between phases. The learned motor primitives generalize to new situations and tasks. Given this library, the robot uses a value function approach to learn a high-level policy for sequencing the motor primitives. The proposed method was successfully evaluated on a real robot performing a bimanual grasping task.", "Imitation learning has traditionally been applied to learn a single task from demonstrations thereof. The requirement of structured and isolated demonstrations limits the scalability of imitation learning approaches as they are difficult to apply to real-world scenarios, where robots have to be able to execute a multitude of tasks. In this paper, we propose a multi-modal imitation learning framework that is able to segment and imitate skills from unlabelled and unstructured demonstrations by learning skill segmentation and imitation learning jointly. The extensive simulation results indicate that our method can efficiently separate the demonstrations into individual skills and learn to imitate them using a single multi-modal policy.", "Abstract: We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. The method is straightforward to implement, is computationally efficient, has little memory requirements, is invariant to diagonal rescaling of the gradients, and is well suited for problems that are large in terms of data and or parameters. The method is also appropriate for non-stationary objectives and problems with very noisy and or sparse gradients. The hyper-parameters have intuitive interpretations and typically require little tuning. Some connections to related algorithms, on which Adam was inspired, are discussed. We also analyze the theoretical convergence properties of the algorithm and provide a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework. Empirical results demonstrate that Adam works well in practice and compares favorably to other stochastic optimization methods. Finally, we discuss AdaMax, a variant of Adam based on the infinity norm.", "Augmenting an agent's control with useful higher-level behaviors called options can greatly reduce the sample complexity of reinforcement learning, but manually designing options is infeasible in high-dimensional and abstract state spaces. While recent work has proposed several techniques for automated option discovery, they do not scale to multi-level hierarchies and to expressive representations such as deep networks. We present Discovery of Deep Options (DDO), a policy-gradient algorithm that discovers parametrized options from a set of demonstration trajectories, and can be used recursively to discover additional levels of the hierarchy. The scalability of our approach to multi-level hierarchies stems from the decoupling of low-level option discovery from high-level meta-control policy learning, facilitated by under-parametrization of the high level. We demonstrate that using the discovered options to augment the action space of Deep Q-Network agents can accelerate learning by guiding exploration in tasks where random actions are unlikely to reach valuable states. We show that DDO is effective in adding options that accelerate learning in 4 out of 5 Atari RAM environments chosen in our experiments. We also show that DDO can discover structure in robot-assisted surgical videos and kinematics that match expert annotation with 72 accuracy." ] }
1812.01483	2945438069	We introduce Compositional Imitation Learning and Execution (CompILE): a framework for learning reusable, variable-length segments of hierarchically-structured behavior from demonstration data. CompILE uses a novel unsupervised, fully-differentiable sequence segmentation module to learn latent encodings of sequential data that can be re-composed and executed to perform new tasks. Once trained, our model generalizes to sequences of longer length and from environment instances not seen during training. We evaluate CompILE in a challenging 2D multi-task environment and a continuous control task, and show that it can find correct task boundaries and event encodings in an unsupervised manner. Latent codes and associated behavior policies discovered by CompILE can be used by a hierarchical agent, where the high-level policy selects actions in the latent code space, and the low-level, task-specific policies are simply the learned decoders. We found that our CompILE-based agent could learn given only sparse rewards, where agents without task-specific policies struggle.	Various solutions for supervised sequence segmentation or task decomposition exist which require varying degrees of supervision @cite_26 @cite_17 @cite_33 @cite_23 . In terms of two recent examples, @cite_33 assume fully-annotated event boundaries and event descriptions at training time whereas TACO @cite_23 only requires (i.e., supervision on sub-task encodings but not on task boundaries) and solves an alignment problem to find a suitable segmentation.	{ "cite_N": [ "@cite_26", "@cite_33", "@cite_23", "@cite_17" ], "mid": [ "950853366", "2963916161", "2962799400", "2519328139" ], "abstract": [ "This chapter provides the background material and literature review for supervised sequence labelling. Section 2.1 briefly reviews supervised learning in general. Section 2.2 covers the classical, non-sequential framework of supervised pattern classification. Section 2.3 defines supervised sequence labelling, and describes the different classes of sequence labelling task that arise under different assumptions about the label sequences.", "Most natural videos contain numerous events. For example, in a video of a “man playing a piano”, the video might also contain “another man dancing” or “a crowd clapping”. We introduce the task of dense-captioning events, which involves both detecting and describing events in a video. We propose a new model that is able to identify all events in a single pass of the video while simultaneously describing the detected events with natural language. Our model introduces a variant of an existing proposal module that is designed to capture both short as well as long events that span minutes. To capture the dependencies between the events in a video, our model introduces a new captioning module that uses contextual information from past and future events to jointly describe all events. We also introduce ActivityNet Captions, a large-scale benchmark for dense-captioning events. ActivityNet Captions contains 20k videos amounting to 849 video hours with 100k total descriptions, each with its unique start and end time. Finally, we report performances of our model for dense-captioning events, video retrieval and localization.", "", "Object proposals have contributed significantly to recent advances in object understanding in images. Inspired by the success of this approach, we introduce Deep Action Proposals (DAPs), an effective and efficient algorithm for generating temporal action proposals from long videos. We show how to take advantage of the vast capacity of deep learning models and memory cells to retrieve from untrimmed videos temporal segments, which are likely to contain actions. A comprehensive evaluation indicates that our approach outperforms previous work on a large scale action benchmark, runs at 134 FPS making it practical for large-scale scenarios, and exhibits an appealing ability to generalize, i.e. to retrieve good quality temporal proposals of actions unseen in training." ] }
1812.01483	2945438069	We introduce Compositional Imitation Learning and Execution (CompILE): a framework for learning reusable, variable-length segments of hierarchically-structured behavior from demonstration data. CompILE uses a novel unsupervised, fully-differentiable sequence segmentation module to learn latent encodings of sequential data that can be re-composed and executed to perform new tasks. Once trained, our model generalizes to sequences of longer length and from environment instances not seen during training. We evaluate CompILE in a challenging 2D multi-task environment and a continuous control task, and show that it can find correct task boundaries and event encodings in an unsupervised manner. Latent codes and associated behavior policies discovered by CompILE can be used by a hierarchical agent, where the high-level policy selects actions in the latent code space, and the low-level, task-specific policies are simply the learned decoders. We found that our CompILE-based agent could learn given only sparse rewards, where agents without task-specific policies struggle.	Outside of the area of , hierarchical reinforcement learning @cite_16 @cite_24 @cite_8 @cite_13 @cite_31 and in particular the options framework @cite_16 @cite_24 @cite_8 similarly deal with the problem of learning segmentations and representations of behavior, but in a purely generative way. Learning with task sketches has also been addressed in this context @cite_1 .	{ "cite_N": [ "@cite_8", "@cite_1", "@cite_24", "@cite_31", "@cite_16", "@cite_13" ], "mid": [ "", "2553882142", "2963262099", "2963912551", "2109910161", "2963286043" ], "abstract": [ "", "We describe a framework for multitask deep reinforcement learning guided by policy sketches. Sketches annotate each task with a sequence of named subtasks, providing high-level structural relationships among tasks, but not providing the detailed guidance required by previous work on learning policy abstractions for RL (e.g. intermediate rewards, subtask completion signals, or intrinsic motivations). Our approach associates every subtask with its own modular subpolicy, and jointly optimizes over full task-specific policies by tying parameters across shared subpolicies. This optimization is accomplished via a simple decoupled actor–critic training objective that facilitates learning common behaviors from dissimilar reward functions. We evaluate the effectiveness of our approach on a maze navigation game and a 2-D Minecraft-inspired crafting game. Both games feature extremely sparse rewards that can be obtained only after completing a number of high-level subgoals (e.g. escaping from a sequence of locked rooms or collecting and combining various ingredients in the proper order). Experiments illustrate two main advantages of our approach. First, we outperform standard baselines that learn task-specific or shared monolithic policies. Second, our method naturally induces a library of primitive behaviors that can be recombined to rapidly acquire policies for new tasks.", "Learning goal-directed behavior in environments with sparse feedback is a major challenge for reinforcement learning algorithms. One of the key difficulties is insufficient exploration, resulting in an agent being unable to learn robust policies. Intrinsically motivated agents can explore new behavior for their own sake rather than to directly solve external goals. Such intrinsic behaviors could eventually help the agent solve tasks posed by the environment. We present hierarchical-DQN (h-DQN), a framework to integrate hierarchical action-value functions, operating at different temporal scales, with goal-driven intrinsically motivated deep reinforcement learning. A top-level q-value function learns a policy over intrinsic goals, while a lower-level function learns a policy over atomic actions to satisfy the given goals. h-DQN allows for flexible goal specifications, such as functions over entities and relations. This provides an efficient space for exploration in complicated environments. We demonstrate the strength of our approach on two problems with very sparse and delayed feedback: (1) a complex discrete stochastic decision process with stochastic transitions, and (2) the classic ATARI game -'Montezuma's Revenge'.", "Building systems that autonomously create temporal abstractions from data is a key challenge in scaling learning and planning in reinforcement learning. One popular approach for addressing this challenge is the options framework (, 2000). However, only recently in (, 2017) was a policy gradient theorem derived for online learning of general purpose options in an end to end fashion. In this work, we extend previous work on this topic that only focuses on learning a two-level hierarchy including options and primitive actions to enable learning simultaneously at multiple resolutions in time. We achieve this by considering an arbitrarily deep hierarchy of options where high level temporally extended options are composed of lower level options with finer resolutions in time. We extend results from (, 2017) and derive policy gradient theorems for a deep hierarchy of options. Our proposed hierarchical option-critic architecture is capable of learning internal policies, termination conditions, and hierarchical compositions over options without the need for any intrinsic rewards or subgoals. Our empirical results in both discrete and continuous environments demonstrate the efficiency of our framework.", "Learning, planning, and representing knowledge at multiple levels of temporal ab- straction are key, longstanding challenges for AI. In this paper we consider how these challenges can be addressed within the mathematical framework of reinforce- ment learning and Markov decision processes (MDPs). We extend the usual notion of action in this framework to include options—closed-loop policies for taking ac- tion over a period of time. Examples of options include picking up an object, going to lunch, and traveling to a distant city, as well as primitive actions such as mus- cle twitches and joint torques. Overall, we show that options enable temporally abstract knowledge and action to be included in the reinforcement learning frame- work in a natural and general way. In particular, we show that options may be used interchangeably with primitive actions in planning methods such as dynamic pro- gramming and in learning methods such as Q-learning. Formally, a set of options defined over an MDP constitutes a semi-Markov decision process (SMDP), and the theory of SMDPs provides the foundation for the theory of options. However, the most interesting issues concern the interplay between the underlying MDP and the SMDP and are thus beyond SMDP theory. We present results for three such cases: 1) we show that the results of planning with options can be used during execution to interrupt options and thereby perform even better than planned, 2) we introduce new intra-option methods that are able to learn about an option from fragments of its execution, and 3) we propose a notion of subgoal that can be used to improve the options themselves. All of these results have precursors in the existing literature; the contribution of this paper is to establish them in a simpler and more general setting with fewer changes to the existing reinforcement learning framework. In particular, we show that these results can be obtained without committing to (or ruling out) any particular approach to state abstraction, hierarchy, function approximation, or the macro-utility problem.", "Deep reinforcement learning has achieved many impressive results in recent years. However, tasks with sparse rewards or long horizons continue to pose significant challenges. To tackle these important problems, we propose a general framework that first learns useful skills in a pre-training environment, and then leverages the acquired skills for learning faster in downstream tasks. Our approach brings together some of the strengths of intrinsic motivation and hierarchical methods: the learning of useful skill is guided by a single proxy reward, the design of which requires very minimal domain knowledge about the downstream tasks. Then a high-level policy is trained on top of these skills, providing a significant improvement of the exploration and allowing to tackle sparse rewards in the downstream tasks. To efficiently pre-train a large span of skills, we use Stochastic Neural Networks combined with an information-theoretic regularizer. Our experiments show that this combination is effective in learning a wide span of interpretable skills in a sample-efficient way, and can significantly boost the learning performance uniformly across a wide range of downstream tasks." ] }
1812.01483	2945438069	We introduce Compositional Imitation Learning and Execution (CompILE): a framework for learning reusable, variable-length segments of hierarchically-structured behavior from demonstration data. CompILE uses a novel unsupervised, fully-differentiable sequence segmentation module to learn latent encodings of sequential data that can be re-composed and executed to perform new tasks. Once trained, our model generalizes to sequences of longer length and from environment instances not seen during training. We evaluate CompILE in a challenging 2D multi-task environment and a continuous control task, and show that it can find correct task boundaries and event encodings in an unsupervised manner. Latent codes and associated behavior policies discovered by CompILE can be used by a hierarchical agent, where the high-level policy selects actions in the latent code space, and the low-level, task-specific policies are simply the learned decoders. We found that our CompILE-based agent could learn given only sparse rewards, where agents without task-specific policies struggle.	Unsupervised segmentation and encoding of sequence data is a similarly important problem in natural language or speech processing, e.g., in the context of word or phoneme segmentation @cite_14 @cite_34 @cite_18 , or in the segmentation of sequential activity data @cite_32 @cite_19 .	{ "cite_N": [ "@cite_18", "@cite_14", "@cite_32", "@cite_19", "@cite_34" ], "mid": [ "", "2126377586", "2962695963", "2753034091", "2530486890" ], "abstract": [ "", "Since the experiments of [Saffran, J., Aslin, R., & Newport, E. (1996). Statistical learning in 8-month-old infants. Science, 274, 1926–1928], there has been a great deal of interest in the question of how statistical regularities in the speech stream might be used by infants to begin to identify individual words. In this work, we use computational modeling to explore the effects of different assumptions the learner might make regarding the nature of words – in particular, how these assumptions affect the kinds of words that are segmented from a corpus of transcribed child-directed speech. We develop several models within a Bayesian ideal observer framework, and use them to examine the consequences of assuming either that words are independent units, or units that help to predict other units. We show through empirical and theoretical results that the assumption of independence causes the learner to undersegment the corpus, with many two- and three-word sequences (e.g. what’s that, do you, in the house) misidentified as individual words. In contrast, when the learner assumes that words are predictive, the resulting segmentation is far more accurate. These results indicate that taking context into account is important for a statistical word segmentation strategy to be successful, and raise the possibility that even young infants may be able to exploit more subtle statistical patterns than have usually been considered.", "We propose a general modeling and inference framework that combines the complementary strengths of probabilistic graphical models and deep learning methods. Our model family composes latent graphical models with neural network observation likelihoods. For inference, we use recognition networks to produce local evidence potentials, then combine them with the model distribution using efficient message-passing algorithms. All components are trained simultaneously with a single stochastic variational inference objective. We illustrate this framework by automatically segmenting and categorizing mouse behavior from raw depth video, and demonstrate several other example models.", "Segmentation and labeling of high dimensional time series data has wide applications in behavior understanding and medical diagnosis. Due to the difficulty in obtaining the label information for high dimensional data, realizing this objective in an unsupervised way is highly desirable. Hidden Semi-Markov Model (HSMM) is a classical tool for this problem. However, existing HSMM and its variants has simple conditional assumptions of observations, thus the ability to capture the nonlinear and complex dynamics within segments is limited. To tackle this limitation, we propose to incorporate the Recurrent Neural Network (RNN) to model the generative process in HSMM, resulting the Recurrent HSMM (R-HSMM). To accelerate the inference while preserving accuracy, we designed a structure encoding function to mimic the exact inference. By generalizing the penalty method to distribution space, we are able to train the model and the encoding function simultaneously. Empirical results show that the proposed R-HSMM achieves the state-of-the-art performances on both synthetic and real-world datasets.", "Sequence-to-sequence models rely on a fixed decomposition of the target sequences into a sequence of tokens that may be words, word-pieces or characters. The choice of these tokens and the decomposition of the target sequences into a sequence of tokens is often static, and independent of the input, output data domains. This can potentially lead to a sub-optimal choice of token dictionaries, as the decomposition is not informed by the particular problem being solved. In this paper we present Latent Sequence Decompositions (LSD), a framework in which the decomposition of sequences into constituent tokens is learnt during the training of the model. The decomposition depends both on the input sequence and on the output sequence. In LSD, during training, the model samples decompositions incrementally, from left to right by locally sampling between valid extensions. We experiment with the Wall Street Journal speech recognition task. Our LSD model achieves 12.9 WER compared to a character baseline of 14.8 WER. When combined with a convolutional network on the encoder, we achieve a WER of 9.6 ." ] }
1812.01600	2903269730	This paper describes AutoFocus, an efficient multi-scale inference algorithm for deep-learning based object detectors. Instead of processing an entire image pyramid, AutoFocus adopts a coarse to fine approach and only processes regions which are likely to contain small objects at finer scales. This is achieved by predicting category agnostic segmentation maps for small objects at coarser scales, called FocusPixels. FocusPixels can be predicted with high recall, and in many cases, they only cover a small fraction of the entire image. To make efficient use of FocusPixels, an algorithm is proposed which generates compact rectangular FocusChips which enclose FocusPixels. The detector is only applied inside FocusChips, which reduces computation while processing finer scales. Different types of error can arise when detections from FocusChips of multiple scales are combined, hence techniques to correct them are proposed. AutoFocus obtains an mAP of 47.9 (68.3 at 50 overlap) on the COCO test-dev set while processing 6.4 images per second on a Titan X (Pascal) GPU. This is 2.5X faster than our multi-scale baseline detector and matches its mAP. The number of pixels processed in the pyramid can be reduced by 5X with a 1 drop in mAP. AutoFocus obtains more than 10 mAP gain compared to RetinaNet but runs at the same speed with the same ResNet-101 backbone.	Image pyramids @cite_24 and convolutional neural networks @cite_25 are fundamental building blocks in the computer vision pipeline. Unfortunately, convolutional neural networks are not scale invariant. Therefore, for instance-level visual recognition problems, to recognize objects of different sizes, it is beneficial to rely on image pyramids @cite_9 . While efficient training solutions have been proposed for multi-scale training @cite_44 , inference on image pyramids remains a computational bottleneck which prohibits their use in practice. Recently, a few methods have been proposed to accelerate multi-scale inference, but they have only been evaluated under constrained settings like pedestrian face detection or object detection in videos @cite_60 @cite_36 @cite_61 @cite_53 @cite_62 @cite_4 . In this work, we propose a simple and pragmatic framework to accelerate multi-scale inference for generic object detection which is evaluated on benchmark datasets.	{ "cite_N": [ "@cite_61", "@cite_62", "@cite_4", "@cite_60", "@cite_36", "@cite_9", "@cite_53", "@cite_24", "@cite_44", "@cite_25" ], "mid": [ "2963585656", "2963212638", "2752236330", "2963897760", "2963859714", "2768489488", "2755268668", "2133155955", "2963381188", "2310919327" ], "abstract": [ "High-performance object detection relies on expensive convolutional networks to compute features, often leading to significant challenges in applications, e.g. those that require detecting objects from video streams in real time. The key to this problem is to trade accuracy for efficiency in an effective way, i.e. reducing the computing cost while maintaining competitive performance. To seek a good balance, previous efforts usually focus on optimizing the model architectures. This paper explores an alternative approach, that is, to reallocate the computation over a scale-time space. The basic idea is to perform expensive detection sparsely and propagate the results across both scales and time with substantially cheaper networks, by exploiting the strong correlations among them. Specifically, we present a unified framework that integrates detection, temporal propagation, and across-scale refinement on a Scale-Time Lattice. On this framework, one can explore various strategies to balance performance and cost. Taking advantage of this flexibility, we further develop an adaptive scheme with the detector invoked on demand and thus obtain improved tradeoff. On ImageNet VID dataset, the proposed method can achieve a competitive mAP 79.6 at 20 fps, or 79.0 at 62 fps as a performance speed tradeoff.1", "This paper introduces an online model for object detection in videos designed to run in real-time on low-powered mobile and embedded devices. Our approach combines fast single-image object detection with convolutional long short term memory (LSTM) layers to create an inter-weaved recurrent-convolutional architecture. Additionally, we propose an efficient Bottleneck-LSTM layer that significantly reduces computational cost compared to regular LSTMs. Our network achieves temporal awareness by using Bottleneck-LSTMs to refine and propagate feature maps across frames. This approach is substantially faster than existing detection methods in video, outperforming the fastest single-frame models in model size and computational cost while attaining accuracy comparable to much more expensive single-frame models on the Imagenet VID 2015 dataset. Our model reaches a real-time inference speed of up to 15 FPS on a mobile CPU.", "Recent advances in computer vision---in the form of deep neural networks---have made it possible to query increasing volumes of video data with high accuracy. However, neural network inference is computationally expensive at scale: applying a state-of-the-art object detector in real time (i.e., 30+ frames per second) to a single video requires a $4000 GPU. In response, we present NoScope, a system for querying videos that can reduce the cost of neural network video analysis by up to three orders of magnitude via inference-optimized model search. Given a target video, object to detect, and reference neural network, NoScope automatically searches for and trains a sequence, or cascade, of models that preserves the accuracy of the reference network but is specialized to the target video and are therefore far less computationally expensive. NoScope cascades two types of models: specialized models that forego the full generality of the reference model but faithfully mimic its behavior for the target video and object; and difference detectors that highlight temporal differences across frames. We show that the optimal cascade architecture differs across videos and objects, so NoScope uses an efficient cost-based optimizer to search across models and cascades. With this approach, NoScope achieves two to three order of magnitude speed-ups (265-15,500x real-time) on binary classification tasks over fixed-angle webcam and surveillance video while maintaining accuracy within 1--5 of state-of-the-art neural networks.", "We introduce a generic framework that reduces the computational cost of object detection while retaining accuracy for scenarios where objects with varied sizes appear in high resolution images. Detection progresses in a coarse-to-fine manner, first on a down-sampled version of the image and then on a sequence of higher resolution regions identified as likely to improve the detection accuracy. Built upon reinforcement learning, our approach consists of a model (R-net) that uses coarse detection results to predict the potential accuracy gain for analyzing a region at a higher resolution and another model (Q-net) that sequentially selects regions to zoom in. Experiments on the Caltech Pedestrians dataset show that our approach reduces the number of processed pixels by over 50 without a drop in detection accuracy. The merits of our approach become more significant on a high resolution test set collected from YFCC100M dataset, where our approach maintains high detection performance while reducing the number of processed pixels by about 70 and the detection time by over 50 .", "Fully convolutional neural network (FCN) has been dominating the game of face detection task for a few years with its congenital capability of sliding-window-searching with shared kernels, which boiled down all the redundant calculation, and most recent state-of-the-art methods such as Faster-RCNN, SSD, YOLO and FPN use FCN as their backbone. So here comes one question: Can we find a universal strategy to further accelerate FCN with higher accuracy, so could accelerate all the recent FCN-based methods? To analyze this, we decompose the face searching space into two orthogonal directions, 'scale' and 'spatial'. Only a few coordinates in the space expanded by the two base vectors indicate foreground. So if FCN could ignore most of the other points, the searching space and false alarm should be significantly boiled down. Based on this philosophy, a novel method named scale estimation and spatial attention proposal (S2AP) is proposed to pay attention to some specific scales in image pyramid and valid locations in each scales layer. Furthermore, we adopt a masked-convolution operation based on the attention result to accelerate FCN calculation. Experiments show that FCN-based method RPN can be accelerated by about 4A— with the help of S2AP and masked-FCN and at the same time it can also achieve the state-of-the-art on FDDB, AFW and MALF face detection benchmarks as well.", "An analysis of different techniques for recognizing and detecting objects under extreme scale variation is presented. Scale specific and scale invariant design of detectors are compared by training them with different configurations of input data. By evaluating the performance of different network architectures for classifying small objects on ImageNet, we show that CNNs are not robust to changes in scale. Based on this analysis, we propose to train and test detectors on the same scales of an image-pyramid. Since small and large objects are difficult to recognize at smaller and larger scales respectively, we present a novel training scheme called Scale Normalization for Image Pyramids (SNIP) which selectively back-propagates the gradients of object instances of different sizes as a function of the image scale. On the COCO dataset, our single model performance is 45.7 and an ensemble of 3 networks obtains an mAP of 48.3 . We use off-the-shelf ImageNet-1000 pre-trained models and only train with bounding box supervision. Our submission won the Best Student Entry in the COCO 2017 challenge. Code will be made available at http: bit.ly 2yXVg4c.", "Object detection is considered one of the most challenging problems in this field of computer vision, as it involves the combination of object classification and object localization within a scene. Recently, deep neural networks (DNNs) have been demonstrated to achieve superior object detection performance compared to other approaches, with YOLOv2 (an improved You Only Look Once model) being one of the state-of-the-art in DNN-based object detection methods in terms of both speed and accuracy. Although YOLOv2 can achieve real-time performance on a powerful GPU, it still remains very challenging for leveraging this approach for real-time object detection in video on embedded computing devices with limited computational power and limited memory. In this paper, we propose a new framework called Fast YOLO, a fast You Only Look Once framework which accelerates YOLOv2 to be able to perform object detection in video on embedded devices in a real-time manner. First, we leverage the evolutionary deep intelligence framework to evolve the YOLOv2 network architecture and produce an optimized architecture (referred to as O-YOLOv2 here) that has 2.8X fewer parameters with just a 2 IOU drop. To further reduce power consumption on embedded devices while maintaining performance, a motion-adaptive inference method is introduced into the proposed Fast YOLO framework to reduce the frequency of deep inference with O-YOLOv2 based on temporal motion characteristics. Experimental results show that the proposed Fast YOLO framework can reduce the number of deep inferences by an average of 38.13 , and an average speedup of 3.3X for objection detection in video compared to the original YOLOv2, leading Fast YOLO to run an average of 18FPS on a Nvidia Jetson TX1 embedded system.", "The extrema in a signal and its first few derivatives provide a useful general purpose qualitative description for many kinds of signals. A fundamental problem in computing such descriptions is scale: a derivative must be taken over some neighborhood, but there is seldom a principled basis for choosing its size. Scale-space filtering is a method that describes signals qualitatively, managing the ambiguity of scale in an organized and natural way. The signal is first expanded by convolution with gaussian masks over a continuum of sizes. This \"scale-space\" image is then collapsed, using its qualitative structure, into a tree providing a concise but complete qualitative description covering all scales of observation. The description is further refined by applying a stability criterion, to identify events that persist of large changes in scale.", "We present SNIPER, an algorithm for performing scale invariant training in instance level visual recognition tasks. Instead of processing every pixel in an image pyramid, SNIPER only processes context regions around ground-truth instances (referred to as chips) at the appropriate scale. For background sampling, these context-regions are generated using proposals extracted from a region proposal network trained with a short learning schedule. Hence, the number of chips generated per image during training adaptively changes based on the scene complexity. SNIPER only processes 30 more pixels compared to the commonly used single scale training at 800x1333 pixels on the COCO dataset. But, it also observes samples from extreme resolutions of the image pyramid, like 1400x2000 pixels. As SNIPER operates on low resolution chips (512x512 pixels), it can have a batch size as large as 20 on a single GPU even with a ResNet-101 backbone. Therefore it can benefit from batch-normalization during training without the need for synchronizing batch-normalization statistics across GPUs. SNIPER brings training of instance level recognition tasks like object detection closer to the protocol for image classification and suggests that the commonly accepted guideline that it is important to train on high resolution images for instance level visual recognition tasks might not be correct. Our implementation based on Faster-RCNN with a ResNet-101 backbone obtains an mAP of 47.6 on the COCO dataset for bounding box detection and can process 5 images per second with a single GPU.", "" ] }
1812.01600	2903269730	This paper describes AutoFocus, an efficient multi-scale inference algorithm for deep-learning based object detectors. Instead of processing an entire image pyramid, AutoFocus adopts a coarse to fine approach and only processes regions which are likely to contain small objects at finer scales. This is achieved by predicting category agnostic segmentation maps for small objects at coarser scales, called FocusPixels. FocusPixels can be predicted with high recall, and in many cases, they only cover a small fraction of the entire image. To make efficient use of FocusPixels, an algorithm is proposed which generates compact rectangular FocusChips which enclose FocusPixels. The detector is only applied inside FocusChips, which reduces computation while processing finer scales. Different types of error can arise when detections from FocusChips of multiple scales are combined, hence techniques to correct them are proposed. AutoFocus obtains an mAP of 47.9 (68.3 at 50 overlap) on the COCO test-dev set while processing 6.4 images per second on a Titan X (Pascal) GPU. This is 2.5X faster than our multi-scale baseline detector and matches its mAP. The number of pixels processed in the pyramid can be reduced by 5X with a 1 drop in mAP. AutoFocus obtains more than 10 mAP gain compared to RetinaNet but runs at the same speed with the same ResNet-101 backbone.	AutoFocus alleviates this problem to a large extent and is designed to provide a smooth trade-off between speed and accuracy. It shows that it is possible to predict the presence of a small object at a coarser scale (referred to as FocusPixels) which enables avoiding computation in large regions of the image at finer scales. These are different from object proposals @cite_20 @cite_31 @cite_7 where region candidates need to have a tight overlap with objects. Learning to predict FocusPixels is an easier task and does not require instance-level reasoning. AutoFocus shares the motivation with saliency based methods which perform a guided search while processing images @cite_33 @cite_5 @cite_58 @cite_54 , but it is designed to predict small objects in coarser scales and they need not be salient.	{ "cite_N": [ "@cite_33", "@cite_7", "@cite_54", "@cite_5", "@cite_31", "@cite_58", "@cite_20" ], "mid": [ "2128272608", "2613718673", "2964210614", "2146103513", "1930392762", "1996326832", "2017691720" ], "abstract": [ "A visual attention system, inspired by the behavior and the neuronal architecture of the early primate visual system, is presented. Multiscale image features are combined into a single topographical saliency map. A dynamical neural network then selects attended locations in order of decreasing saliency. The system breaks down the complex problem of scene understanding by rapidly selecting, in a computationally efficient manner, conspicuous locations to be analyzed in detail.", "State-of-the-art object detection networks depend on region proposal algorithms to hypothesize object locations. Advances like SPPnet [7] and Fast R-CNN [5] have reduced the running time of these detection networks, exposing region proposal computation as a bottleneck. In this work, we introduce a Region Proposal Network (RPN) that shares full-image convolutional features with the detection network, thus enabling nearly cost-free region proposals. An RPN is a fully-convolutional network that simultaneously predicts object bounds and objectness scores at each position. RPNs are trained end-to-end to generate high-quality region proposals, which are used by Fast R-CNN for detection. With a simple alternating optimization, RPN and Fast R-CNN can be trained to share convolutional features. For the very deep VGG-16 model [19], our detection system has a frame rate of 5fps (including all steps) on a GPU, while achieving state-of-the-art object detection accuracy on PASCAL VOC 2007 (73.2 mAP) and 2012 (70.4 mAP) using 300 proposals per image. Code is available at https: github.com ShaoqingRen faster_rcnn.", "In this work, we propose an efficient and effective approach for unconstrained salient object detection in images using deep convolutional neural networks. Instead of generating thousands of candidate bounding boxes and refining them, our network directly learns to generate the saliency map containing the exact number of salient objects. During training, we convert the ground-truth rectangular boxes to Gaussian distributions that better capture the ROI regarding individual salient objects. During inference, the network predicts Gaussian distributions centered at salient objects with an appropriate covariance, from which bounding boxes are easily inferred. Notably, our network performs saliency map prediction without pixel-level annotations, salient object detection without object proposals, and salient object subitizing simultaneously, all in a single pass within a unified framework. Extensive experiments show that our approach outperforms existing methods on various datasets by a large margin, and achieves more than 100 fps with VGG16 network on a single GPU during inference.", "The ability of human visual system to detect visual saliency is extraordinarily fast and reliable. However, computational modeling of this basic intelligent behavior still remains a challenge. This paper presents a simple method for the visual saliency detection. Our model is independent of features, categories, or other forms of prior knowledge of the objects. By analyzing the log-spectrum of an input image, we extract the spectral residual of an image in spectral domain, and propose a fast method to construct the corresponding saliency map in spatial domain. We test this model on both natural pictures and artificial images such as psychological patterns. The result indicate fast and robust saliency detection of our method.", "Object class detectors typically apply a window classifier to all the windows in a large set, either in a sliding window manner or using object proposals. In this paper, we develop an active search strategy that sequentially chooses the next window to evaluate based on all the information gathered before. This results in a substantial reduction in the number of classifier evaluations and in a more elegant approach in general. Our search strategy is guided by two forces. First, we exploit context as the statistical relation between the appearance of a window and its location relative to the object, as observed in the training set. This enables to jump across distant regions in the image (e.g. observing a sky region suggests that cars might be far below) and is done efficiently in a Random Forest framework. Second, we exploit the score of the classifier to attract the search to promising areas surrounding a highly scored window, and to keep away from areas near low scored ones. Our search strategy can be applied on top of any classifier as it treats it as a black-box. In experiments with R-CNN on the challenging SUN2012 dataset, our method matches the detection accuracy of evaluating all windows independently, while evaluating 9× fewer windows.", "In this paper, we study the salient object detection problem for images. We formulate this problem as a binary labeling task where we separate the salient object from the background. We propose a set of novel features, including multiscale contrast, center-surround histogram, and color spatial distribution, to describe a salient object locally, regionally, and globally. A conditional random field is learned to effectively combine these features for salient object detection. Further, we extend the proposed approach to detect a salient object from sequential images by introducing the dynamic salient features. We collected a large image database containing tens of thousands of carefully labeled images by multiple users and a video segment database, and conducted a set of experiments over them to demonstrate the effectiveness of the proposed approach.", "We present a novel framework for generating and ranking plausible objects hypotheses in an image using bottom-up processes and mid-level cues. The object hypotheses are represented as figure-ground segmentations, and are extracted automatically, without prior knowledge about properties of individual object classes, by solving a sequence of constrained parametric min-cut problems (CPMC) on a regular image grid. We then learn to rank the object hypotheses by training a continuous model to predict how plausible the segments are, given their mid-level region properties. We show that this algorithm significantly outperforms the state of the art for low-level segmentation in the VOC09 segmentation dataset. It achieves the same average best segmentation covering as the best performing technique to date [2], 0.61 when using just the top 7 ranked segments, instead of the full hierarchy in [2]. Our method achieves 0.78 average best covering using 154 segments. In a companion paper [18], we also show that the algorithm achieves state-of-the art results when used in a segmentation-based recognition pipeline." ] }
1811.08883	2901394229	We report competitive results on object detection and instance segmentation on the COCO dataset using standard models trained from random initialization. The results are no worse than their ImageNet pre-training counterparts even when using the hyper-parameters of the baseline system (Mask R-CNN) that were optimized for fine-tuning pre-trained models, with the sole exception of increasing the number of training iterations so the randomly initialized models may converge. Training from random initialization is surprisingly robust; our results hold even when: (i) using only 10 of the training data, (ii) for deeper and wider models, and (iii) for multiple tasks and metrics. Experiments show that ImageNet pre-training speeds up convergence early in training, but does not necessarily provide regularization or improve final target task accuracy. To push the envelope we demonstrate 50.9 AP on COCO object detection without using any external data---a result on par with the top COCO 2017 competition results that used ImageNet pre-training. These observations challenge the conventional wisdom of ImageNet pre-training for dependent tasks and we expect these discoveries will encourage people to rethink the current de facto paradigm of pre-training and fine-tuning' in computer vision.	Before the prevalence of the pre-training and fine-tuning' paradigm, object detectors were trained ( , @cite_49 @cite_41 @cite_48 )---a fact that is somewhat overlooked today. In fact, .	{ "cite_N": [ "@cite_41", "@cite_48", "@cite_49" ], "mid": [ "2125713050", "2130306094", "" ], "abstract": [ "We present a neural network-based face detection system. A retinally connected neural network examines small windows of an image, and decides whether each window contains a face. The system arbitrates between multiple networks to improve performance over a single network. We use a bootstrap algorithm for training, which adds false detections into the training set as training progresses. This eliminates the difficult task of manually selecting non-face training examples, which must be chosen to span the entire space of non-face images. Comparisons with another state-of-the-art face detection system are presented; our system has better performance in terms of detection and false-positive rates.", "Deep Neural Networks (DNNs) have recently shown outstanding performance on image classification tasks [14]. In this paper we go one step further and address the problem of object detection using DNNs, that is not only classifying but also precisely localizing objects of various classes. We present a simple and yet powerful formulation of object detection as a regression problem to object bounding box masks. We define a multi-scale inference procedure which is able to produce high-resolution object detections at a low cost by a few network applications. State-of-the-art performance of the approach is shown on Pascal VOC.", "" ] }
1811.08883	2901394229	We report competitive results on object detection and instance segmentation on the COCO dataset using standard models trained from random initialization. The results are no worse than their ImageNet pre-training counterparts even when using the hyper-parameters of the baseline system (Mask R-CNN) that were optimized for fine-tuning pre-trained models, with the sole exception of increasing the number of training iterations so the randomly initialized models may converge. Training from random initialization is surprisingly robust; our results hold even when: (i) using only 10 of the training data, (ii) for deeper and wider models, and (iii) for multiple tasks and metrics. Experiments show that ImageNet pre-training speeds up convergence early in training, but does not necessarily provide regularization or improve final target task accuracy. To push the envelope we demonstrate 50.9 AP on COCO object detection without using any external data---a result on par with the top COCO 2017 competition results that used ImageNet pre-training. These observations challenge the conventional wisdom of ImageNet pre-training for dependent tasks and we expect these discoveries will encourage people to rethink the current de facto paradigm of pre-training and fine-tuning' in computer vision.	Given the success of pre-training in the R-CNN paper @cite_15 , later analysis @cite_29 found that pre-training plays an important role in detector accuracy when training data is limited, but also illustrated that and can achieve 90 As modern object detectors @cite_15 @cite_33 @cite_1 @cite_8 @cite_4 @cite_40 @cite_30 @cite_17 evolved under the pre-training paradigm, the belief that training from scratch is non-trivial became conventional wisdom. Shen al @cite_31 argued for to obtain a detector that is optimized for the accuracy when trained from scratch. They designed a specialized detector driven by deeply supervised networks @cite_22 and dense connections @cite_27 . DetNet @cite_11 and CornerNet @cite_18 also present results when training detectors from scratch. Similar to @cite_31 , these works @cite_11 @cite_18 focus on designing detection-specific architectures. However, in @cite_31 @cite_11 @cite_18 .	{ "cite_N": [ "@cite_30", "@cite_31", "@cite_11", "@cite_4", "@cite_33", "@cite_22", "@cite_8", "@cite_18", "@cite_29", "@cite_1", "@cite_40", "@cite_27", "@cite_15", "@cite_17" ], "mid": [ "2949533892", "2743388417", "2797527871", "", "2179352600", "", "", "2886335102", "2160921898", "", "2193145675", "2511730936", "2102605133", "" ], "abstract": [ "Feature pyramids are a basic component in recognition systems for detecting objects at different scales. But recent deep learning object detectors have avoided pyramid representations, in part because they are compute and memory intensive. In this paper, we exploit the inherent multi-scale, pyramidal hierarchy of deep convolutional networks to construct feature pyramids with marginal extra cost. A top-down architecture with lateral connections is developed for building high-level semantic feature maps at all scales. This architecture, called a Feature Pyramid Network (FPN), shows significant improvement as a generic feature extractor in several applications. Using FPN in a basic Faster R-CNN system, our method achieves state-of-the-art single-model results on the COCO detection benchmark without bells and whistles, surpassing all existing single-model entries including those from the COCO 2016 challenge winners. In addition, our method can run at 5 FPS on a GPU and thus is a practical and accurate solution to multi-scale object detection. Code will be made publicly available.", "We present Deeply Supervised Object Detector (DSOD), a framework that can learn object detectors from scratch. State-of-the-art object objectors rely heavily on the off-the-shelf networks pre-trained on large-scale classification datasets like ImageNet, which incurs learning bias due to the difference on both the loss functions and the category distributions between classification and detection tasks. Model fine-tuning for the detection task could alleviate this bias to some extent but not fundamentally. Besides, transferring pre-trained models from classification to detection between discrepant domains is even more difficult (e.g. RGB to depth images). A better solution to tackle these two critical problems is to train object detectors from scratch, which motivates our proposed DSOD. Previous efforts in this direction mostly failed due to much more complicated loss functions and limited training data in object detection. In DSOD, we contribute a set of design principles for training object detectors from scratch. One of the key findings is that deep supervision, enabled by dense layer-wise connections, plays a critical role in learning a good detector. Combining with several other principles, we develop DSOD following the single-shot detection (SSD) framework. Experiments on PASCAL VOC 2007, 2012 and MS COCO datasets demonstrate that DSOD can achieve better results than the state-of-the-art solutions with much more compact models. For instance, DSOD outperforms SSD on all three benchmarks with real-time detection speed, while requires only 1 2 parameters to SSD and 1 10 parameters to Faster RCNN. Our code and models are available at: this https URL .", "Recent CNN based object detectors, no matter one-stage methods like YOLO, SSD, and RetinaNe or two-stage detectors like Faster R-CNN, R-FCN and FPN are usually trying to directly finetune from ImageNet pre-trained models designed for image classification. There has been little work discussing on the backbone feature extractor specifically designed for the object detection. More importantly, there are several differences between the tasks of image classification and object detection. 1. Recent object detectors like FPN and RetinaNet usually involve extra stages against the task of image classification to handle the objects with various scales. 2. Object detection not only needs to recognize the category of the object instances but also spatially locate the position. Large downsampling factor brings large valid receptive field, which is good for image classification but compromises the object location ability. Due to the gap between the image classification and object detection, we propose DetNet in this paper, which is a novel backbone network specifically designed for object detection. Moreover, DetNet includes the extra stages against traditional backbone network for image classification, while maintains high spatial resolution in deeper layers. Without any bells and whistles, state-of-the-art results have been obtained for both object detection and instance segmentation on the MSCOCO benchmark based on our DetNet (4.8G FLOPs) backbone. The code will be released for the reproduction.", "", "Existing deep convolutional neural networks (CNNs) require a fixed-size (e.g. 224×224) input image. This requirement is “artificial” and may hurt the recognition accuracy for the images or sub-images of an arbitrary size scale. In this work, we equip the networks with a more principled pooling strategy, “spatial pyramid pooling”, to eliminate the above requirement. The new network structure, called SPP-net, can generate a fixed-length representation regardless of image size scale. By removing the fixed-size limitation, we can improve all CNN-based image classification methods in general. Our SPP-net achieves state-of-the-art accuracy on the datasets of ImageNet 2012, Pascal VOC 2007, and Caltech101.", "", "", "We propose CornerNet, a new approach to object detection where we detect an object bounding box as a pair of keypoints, the top-left corner and the bottom-right corner, using a single convolution neural network. By detecting objects as paired keypoints, we eliminate the need for designing a set of anchor boxes commonly used in prior single-stage detectors. In addition to our novel formulation, we introduce corner pooling, a new type of pooling layer that helps the network better localize corners. Experiments show that CornerNet achieves a 42.2 AP on MS COCO, outperforming all existing one-stage detectors.", "In the last two years, convolutional neural networks (CNNs) have achieved an impressive suite of results on standard recognition datasets and tasks. CNN-based features seem poised to quickly replace engineered representations, such as SIFT and HOG. However, compared to SIFT and HOG, we understand much less about the nature of the features learned by large CNNs. In this paper, we experimentally probe several aspects of CNN feature learning in an attempt to help practitioners gain useful, evidence-backed intuitions about how to apply CNNs to computer vision problems.", "", "We present a method for detecting objects in images using a single deep neural network. Our approach, named SSD, discretizes the output space of bounding boxes into a set of default boxes over different aspect ratios and scales per feature map location. At prediction time, the network generates scores for the presence of each object category in each default box and produces adjustments to the box to better match the object shape. Additionally, the network combines predictions from multiple feature maps with different resolutions to naturally handle objects of various sizes. SSD is simple relative to methods that require object proposals because it completely eliminates proposal generation and subsequent pixel or feature resampling stages and encapsulates all computation in a single network. This makes SSD easy to train and straightforward to integrate into systems that require a detection component. Experimental results on the PASCAL VOC, COCO, and ILSVRC datasets confirm that SSD has competitive accuracy to methods that utilize an additional object proposal step and is much faster, while providing a unified framework for both training and inference. For (300 300 ) input, SSD achieves 74.3 mAP on VOC2007 test at 59 FPS on a Nvidia Titan X and for (512 512 ) input, SSD achieves 76.9 mAP, outperforming a comparable state of the art Faster R-CNN model. Compared to other single stage methods, SSD has much better accuracy even with a smaller input image size. Code is available at https: github.com weiliu89 caffe tree ssd.", "Recent work has shown that convolutional networks can be substantially deeper, more accurate, and efficient to train if they contain shorter connections between layers close to the input and those close to the output. In this paper, we embrace this observation and introduce the Dense Convolutional Network (DenseNet), which connects each layer to every other layer in a feed-forward fashion. Whereas traditional convolutional networks with L layers have L connections - one between each layer and its subsequent layer - our network has L(L+1) 2 direct connections. For each layer, the feature-maps of all preceding layers are used as inputs, and its own feature-maps are used as inputs into all subsequent layers. DenseNets have several compelling advantages: they alleviate the vanishing-gradient problem, strengthen feature propagation, encourage feature reuse, and substantially reduce the number of parameters. We evaluate our proposed architecture on four highly competitive object recognition benchmark tasks (CIFAR-10, CIFAR-100, SVHN, and ImageNet). DenseNets obtain significant improvements over the state-of-the-art on most of them, whilst requiring less computation to achieve high performance. Code and pre-trained models are available at this https URL .", "Object detection performance, as measured on the canonical PASCAL VOC dataset, has plateaued in the last few years. The best-performing methods are complex ensemble systems that typically combine multiple low-level image features with high-level context. In this paper, we propose a simple and scalable detection algorithm that improves mean average precision (mAP) by more than 30 relative to the previous best result on VOC 2012 -- achieving a mAP of 53.3 . Our approach combines two key insights: (1) one can apply high-capacity convolutional neural networks (CNNs) to bottom-up region proposals in order to localize and segment objects and (2) when labeled training data is scarce, supervised pre-training for an auxiliary task, followed by domain-specific fine-tuning, yields a significant performance boost. Since we combine region proposals with CNNs, we call our method R-CNN: Regions with CNN features. We also present experiments that provide insight into what the network learns, revealing a rich hierarchy of image features. Source code for the complete system is available at http: www.cs.berkeley.edu rbg rcnn.", "" ] }
1811.08585	2901467893	Unsupervised domain adaptation (UDA) transfers knowledge from a label-rich source domain to a fully-unlabeled target domain. To tackle this task, recent approaches resort to discriminative domain transfer in virtue of pseudo-labels to enforce the class-level distribution alignment across the source and target domains. These methods, however, are vulnerable to the error accumulation and thus incapable of preserving cross-domain category consistency, as the pseudo-labeling accuracy is not guaranteed explicitly. In this paper, we propose the Progressive Feature Alignment Network (PFAN) to align the discriminative features across domains progressively and effectively, via exploiting the intra-class variation in the target domain. To be specific, we first develop an Easy-to-Hard Transfer Strategy (EHTS) and an Adaptive Prototype Alignment (APA) step to train our model iteratively and alternatively. Moreover, upon observing that a good domain adaptation usually requires a non-saturated source classifier, we consider a simple yet efficient way to retard the convergence speed of the source classification loss by further involving a temperature variate into the soft-max function. The extensive experimental results reveal that the proposed PFAN exceeds the state-of-the-art performance on three UDA datasets.	Inspired by the recent success of generative adversarial networks (GAN) @cite_16 , deep adversarial domain adaptation has received increasing attention in learning domain-invariant representations to reduce the domain discrepancy and provide remarkable results @cite_6 @cite_7 @cite_19 @cite_29 @cite_13 @cite_26 . These methods try to find a feature space such that confusion between the source and the target distributions in that space is maximal. For example, @cite_6 proposed a gradient reversal layer to train a feature extractor that produces features which maximize the domain binary classifier loss, while simultaneously minimizing the label predictor loss.	{ "cite_N": [ "@cite_26", "@cite_7", "@cite_29", "@cite_6", "@cite_19", "@cite_16", "@cite_13" ], "mid": [ "2964055354", "2593768305", "2962986791", "2963826681", "2788768841", "2099471712", "2798681837" ], "abstract": [ "In this paper, we make two contributions to unsupervised domain adaptation (UDA) using the convolutional neural network (CNN). First, our approach transfers knowledge in all the convolutional layers through attention alignment. Most previous methods align high-level representations, e.g., activations of the fully connected (FC) layers. In these methods, however, the convolutional layers which underpin critical low-level domain knowledge cannot be updated directly towards reducing domain discrepancy. Specifically, we assume that the discriminative regions in an image are relatively invariant to image style changes. Based on this assumption, we propose an attention alignment scheme on all the target convolutional layers to uncover the knowledge shared by the source domain. Second, we estimate the posterior label distribution of the unlabeled data for target network training. Previous methods, which iteratively update the pseudo labels by the target network and refine the target network by the updated pseudo labels, are vulnerable to label estimation errors. Instead, our approach uses category distribution to calculate the cross-entropy loss for training, thereby ameliorating the error accumulation of the estimated labels. The two contributions allow our approach to outperform the state-of-the-art methods by +2.6 on the Office-31 dataset.", "Adversarial learning methods are a promising approach to training robust deep networks, and can generate complex samples across diverse domains. They can also improve recognition despite the presence of domain shift or dataset bias: recent adversarial approaches to unsupervised domain adaptation reduce the difference between the training and test domain distributions and thus improve generalization performance. However, while generative adversarial networks (GANs) show compelling visualizations, they are not optimal on discriminative tasks and can be limited to smaller shifts. On the other hand, discriminative approaches can handle larger domain shifts, but impose tied weights on the model and do not exploit a GAN-based loss. In this work, we first outline a novel generalized framework for adversarial adaptation, which subsumes recent state-of-the-art approaches as special cases, and use this generalized view to better relate prior approaches. We then propose a previously unexplored instance of our general framework which combines discriminative modeling, untied weight sharing, and a GAN loss, which we call Adversarial Discriminative Domain Adaptation (ADDA). We show that ADDA is more effective yet considerably simpler than competing domain-adversarial methods, and demonstrate the promise of our approach by exceeding state-of-the-art unsupervised adaptation results on standard domain adaptation tasks as well as a difficult cross-modality object classification task.", "This paper proposes an importance weighted adversarial nets-based method for unsupervised domain adaptation, specific for partial domain adaptation where the target domain has less number of classes compared to the source domain. Previous domain adaptation methods generally assume the identical label spaces, such that reducing the distribution divergence leads to feasible knowledge transfer. However, such an assumption is no longer valid in a more realistic scenario that requires adaptation from a larger and more diverse source domain to a smaller target domain with less number of classes. This paper extends the adversarial nets-based domain adaptation and proposes a novel adversarial nets-based partial domain adaptation method to identify the source samples that are potentially from the outlier classes and, at the same time, reduce the shift of shared classes between domains.", "Top-performing deep architectures are trained on massive amounts of labeled data. In the absence of labeled data for a certain task, domain adaptation often provides an attractive option given that labeled data of similar nature but from a different domain (e.g. synthetic images) are available. Here, we propose a new approach to domain adaptation in deep architectures that can be trained on large amount of labeled data from the source domain and large amount of unlabeled data from the target domain (no labeled target-domain data is necessary). As the training progresses, the approach promotes the emergence of \"deep\" features that are (i) discriminative for the main learning task on the source domain and (ii) invariant with respect to the shift between the domains. We show that this adaptation behaviour can be achieved in almost any feed-forward model by augmenting it with few standard layers and a simple new gradient reversal layer. The resulting augmented architecture can be trained using standard back propagation. Overall, the approach can be implemented with little effort using any of the deep-learning packages. The method performs very well in a series of image classification experiments, achieving adaptation effect in the presence of big domain shifts and outperforming previous state-of-the-art on Office datasets.", "", "We propose a new framework for estimating generative models via an adversarial process, in which we simultaneously train two models: a generative model G that captures the data distribution, and a discriminative model D that estimates the probability that a sample came from the training data rather than G. The training procedure for G is to maximize the probability of D making a mistake. This framework corresponds to a minimax two-player game. In the space of arbitrary functions G and D, a unique solution exists, with G recovering the training data distribution and D equal to ½ everywhere. In the case where G and D are defined by multilayer perceptrons, the entire system can be trained with backpropagation. There is no need for any Markov chains or unrolled approximate inference networks during either training or generation of samples. Experiments demonstrate the potential of the framework through qualitative and quantitative evaluation of the generated samples.", "In this paper, we propose a new unsupervised domain adaptation approach called Collaborative and Adversarial Network (CAN) through domain-collaborative and domain-adversarial training of neural networks. We add several domain classifiers on multiple CNN feature extraction blocks1, in which each domain classifier is connected to the hidden representations from one block and one loss function is defined based on the hidden presentation and the domain labels (e.g., source and target). We design a new loss function by integrating the losses from all blocks in order to learn domain informative representations from lower blocks through collaborative learning and learn domain uninformative representations from higher blocks through adversarial learning. We further extend our CAN method as Incremental CAN (iCAN), in which we iteratively select a set of pseudo-labelled target samples based on the image classifier and the last domain classifier from the previous training epoch and re-train our CAN model by using the enlarged training set. Comprehensive experiments on two benchmark datasets Office and ImageCLEF-DA clearly demonstrate the effectiveness of our newly proposed approaches CAN and iCAN for unsupervised domain adaptation." ] }
1811.08585	2901467893	Unsupervised domain adaptation (UDA) transfers knowledge from a label-rich source domain to a fully-unlabeled target domain. To tackle this task, recent approaches resort to discriminative domain transfer in virtue of pseudo-labels to enforce the class-level distribution alignment across the source and target domains. These methods, however, are vulnerable to the error accumulation and thus incapable of preserving cross-domain category consistency, as the pseudo-labeling accuracy is not guaranteed explicitly. In this paper, we propose the Progressive Feature Alignment Network (PFAN) to align the discriminative features across domains progressively and effectively, via exploiting the intra-class variation in the target domain. To be specific, we first develop an Easy-to-Hard Transfer Strategy (EHTS) and an Adaptive Prototype Alignment (APA) step to train our model iteratively and alternatively. Moreover, upon observing that a good domain adaptation usually requires a non-saturated source classifier, we consider a simple yet efficient way to retard the convergence speed of the source classification loss by further involving a temperature variate into the soft-max function. The extensive experimental results reveal that the proposed PFAN exceeds the state-of-the-art performance on three UDA datasets.	Many approaches utilize a distance metric to measure the domain discrepancy between the source and target domains, such as maximum mean discrepancy (MMD), KL-divergence or Wasserstein distance @cite_31 @cite_5 @cite_38 @cite_3 @cite_9 @cite_34 . Most prior efforts intend to achieve domain alignment by matching @math and @math . However, an exact domain-level alignment does not imply a fine-grained class-to-class overlap. Thus, it is important to pursue the category-level alignment under the absence of target true labels.	{ "cite_N": [ "@cite_38", "@cite_9", "@cite_3", "@cite_5", "@cite_31", "@cite_34" ], "mid": [ "2964288524", "2964285681", "2279034837", "2159291411", "2212660284", "2798377719" ], "abstract": [ "Deep neural networks are able to learn powerful representations from large quantities of labeled input data, however they cannot always generalize well across changes in input distributions. Domain adaptation algorithms have been proposed to compensate for the degradation in performance due to domain shift. In this paper, we address the case when the target domain is unlabeled, requiring unsupervised adaptation. CORAL [18] is a simple unsupervised domain adaptation method that aligns the second-order statistics of the source and target distributions with a linear transformation. Here, we extend CORAL to learn a nonlinear transformation that aligns correlations of layer activations in deep neural networks (Deep CORAL). Experiments on standard benchmark datasets show state-of-the-art performance. Our code is available at: https: github.com VisionLearningGroup CORAL.", "In domain adaptation, maximum mean discrepancy (MMD) has been widely adopted as a discrepancy metric between the distributions of source and target domains. However, existing MMD-based domain adaptation methods generally ignore the changes of class prior distributions, i.e., class weight bias across domains. This remains an open problem but ubiquitous for domain adaptation, which can be caused by changes in sample selection criteria and application scenarios. We show that MMD cannot account for class weight bias and results in degraded domain adaptation performance. To address this issue, a weighted MMD model is proposed in this paper. Specifically, we introduce class-specific auxiliary weights into the original MMD for exploiting the class prior probability on source and target domains, whose challenge lies in the fact that the class label in target domain is unavailable. To account for it, our proposed weighted MMD model is defined by introducing an auxiliary weight for each class in the source domain, and a classification EM algorithm is suggested by alternating between assigning the pseudo-labels, estimating auxiliary weights and updating model parameters. Extensive experiments demonstrate the superiority of our weighted MMD over conventional MMD for domain adaptation.", "The recent success of deep neural networks relies on massive amounts of labeled data. For a target task where labeled data is unavailable, domain adaptation can transfer a learner from a different source domain. In this paper, we propose a new approach to domain adaptation in deep networks that can jointly learn adaptive classifiers and transferable features from labeled data in the source domain and unlabeled data in the target domain. We relax a shared-classifier assumption made by previous methods and assume that the source classifier and target classifier differ by a residual function. We enable classifier adaptation by plugging several layers into deep network to explicitly learn the residual function with reference to the target classifier. We fuse features of multiple layers with tensor product and embed them into reproducing kernel Hilbert spaces to match distributions for feature adaptation. The adaptation can be achieved in most feed-forward models by extending them with new residual layers and loss functions, which can be trained efficiently via back-propagation. Empirical evidence shows that the new approach outperforms state of the art methods on standard domain adaptation benchmarks.", "Recent studies reveal that a deep neural network can learn transferable features which generalize well to novel tasks for domain adaptation. However, as deep features eventually transition from general to specific along the network, the feature transferability drops significantly in higher layers with increasing domain discrepancy. Hence, it is important to formally reduce the dataset bias and enhance the transferability in task-specific layers. In this paper, we propose a new Deep Adaptation Network (DAN) architecture, which generalizes deep convolutional neural network to the domain adaptation scenario. In DAN, hidden representations of all task-specific layers are embedded in a reproducing kernel Hilbert space where the mean embeddings of different domain distributions can be explicitly matched. The domain discrepancy is further reduced using an optimal multikernel selection method for mean embedding matching. DAN can learn transferable features with statistical guarantees, and can scale linearly by unbiased estimate of kernel embedding. Extensive empirical evidence shows that the proposed architecture yields state-of-the-art image classification error rates on standard domain adaptation benchmarks.", "We propose a framework for analyzing and comparing distributions, which we use to construct statistical tests to determine if two samples are drawn from different distributions. Our test statistic is the largest difference in expectations over functions in the unit ball of a reproducing kernel Hilbert space (RKHS), and is called the maximum mean discrepancy (MMD).We present two distribution free tests based on large deviation bounds for the MMD, and a third test based on the asymptotic distribution of this statistic. The MMD can be computed in quadratic time, although efficient linear time approximations are available. Our statistic is an instance of an integral probability metric, and various classical metrics on distributions are obtained when alternative function classes are used in place of an RKHS. We apply our two-sample tests to a variety of problems, including attribute matching for databases using the Hungarian marriage method, where they perform strongly. Excellent performance is also obtained when comparing distributions over graphs, for which these are the first such tests.", "Unsupervised Domain Adaptation (UDA) aims to transfer domain knowledge from existing well-defined tasks to new ones where labels are unavailable. In the real-world applications, as the domain (task) discrepancies are usually uncontrollable, it is significantly motivated to match the feature distributions even if the domain discrepancies are disparate. Additionally, as no label is available in the target domain, how to successfully adapt the classifier from the source to the target domain still remains an open question. In this paper, we propose the Re-weighted Adversarial Adaptation Network (RAAN) to reduce the feature distribution divergence and adapt the classifier when domain discrepancies are disparate. Specifically, to alleviate the need of common supports in matching the feature distribution, we choose to minimize optimal transport (OT) based Earth-Mover (EM) distance and reformulate it to a minimax objective function. Utilizing this, RAAN can be trained in an end-to-end and adversarial manner. To further adapt the classifier, we propose to match the label distribution and embed it into the adversarial training. Finally, after extensive evaluation of our method using UDA datasets of varying difficulty, RAAN achieved the state-of-the-art results and outperformed other methods by a large margin when the domain shifts are disparate." ] }
1811.08585	2901467893	Unsupervised domain adaptation (UDA) transfers knowledge from a label-rich source domain to a fully-unlabeled target domain. To tackle this task, recent approaches resort to discriminative domain transfer in virtue of pseudo-labels to enforce the class-level distribution alignment across the source and target domains. These methods, however, are vulnerable to the error accumulation and thus incapable of preserving cross-domain category consistency, as the pseudo-labeling accuracy is not guaranteed explicitly. In this paper, we propose the Progressive Feature Alignment Network (PFAN) to align the discriminative features across domains progressively and effectively, via exploiting the intra-class variation in the target domain. To be specific, we first develop an Easy-to-Hard Transfer Strategy (EHTS) and an Adaptive Prototype Alignment (APA) step to train our model iteratively and alternatively. Moreover, upon observing that a good domain adaptation usually requires a non-saturated source classifier, we consider a simple yet efficient way to retard the convergence speed of the source classification loss by further involving a temperature variate into the soft-max function. The extensive experimental results reveal that the proposed PFAN exceeds the state-of-the-art performance on three UDA datasets.	@cite_30 @cite_23 @cite_25 @cite_35 @cite_14 @cite_13 @cite_36 utilize the pseudo-labels to compensate the lack of categorical information in the target domain. @cite_25 jointly matched both the marginal distribution and conditional distribution using a revised MMD. @cite_14 utilized an asymmetric tri-training strategy to learn discriminative representations for the target domain. @cite_13 iteratively selected pseudo-labeled target samples based on the classifier from the previous training epoch and re-trained the model by using the enlarged training set. @cite_36 proposed to assign pseudo-labels to all target samples and utilize them to achieve semantic alignment across domains. However, these approaches highly relied on the hypothesis that correctly-pseudo-labeled samples can reduce the bias caused by falsely-pseudo-labeled samples. They do not explicitly alleviate those falsely-pseudo-labeled samples. When the falsely-pseudo-labeled samples take the prominent position, their performances will be limited.	{ "cite_N": [ "@cite_30", "@cite_35", "@cite_14", "@cite_36", "@cite_23", "@cite_13", "@cite_25" ], "mid": [ "2159570078", "2551835155", "2594718649", "2803297029", "", "2798681837", "2096943734" ], "abstract": [ "This paper addresses pattern classification in the framework of domain adaptation by considering methods that solve problems in which training data are assumed to be available only for a source domain different (even if related) from the target domain of (unlabeled) test data. Two main novel contributions are proposed: 1) a domain adaptation support vector machine (DASVM) technique which extends the formulation of support vector machines (SVMs) to the domain adaptation framework and 2) a circular indirect accuracy assessment strategy for validating the learning of domain adaptation classifiers when no true labels for the target--domain instances are available. Experimental results, obtained on a series of two-dimensional toy problems and on two real data sets related to brain computer interface and remote sensing applications, confirmed the effectiveness and the reliability of both the DASVM technique and the proposed circular validation strategy.", "Supervised learning with large scale labelled datasets and deep layered models has caused a paradigm shift in diverse areas in learning and recognition. However, this approach still suffers from generalization issues under the presence of a domain shift between the training and the test data distribution. Since unsupervised domain adaptation algorithms directly address this domain shift problem between a labelled source dataset and an unlabelled target dataset, recent papers have shown promising results by fine-tuning the networks with domain adaptation loss functions which try to align the mismatch between the training and testing data distributions. Nevertheless, these recent deep learning based domain adaptation approaches still suffer from issues such as high sensitivity to the gradient reversal hyperparameters and overfitting during the fine-tuning stage. In this paper, we propose a unified deep learning framework where the representation, cross domain transformation, and target label inference are all jointly optimized in an end-to-end fashion for unsupervised domain adaptation. Our experiments show that the proposed method significantly outperforms state-of-the-art algorithms in both object recognition and digit classification experiments by a large margin. We will make our learned models as well as the source code available immediately upon acceptance.", "It is important to apply models trained on a large number of labeled samples to different domains because collecting many labeled samples in various domains is expensive. To learn discriminative representations for the target domain, we assume that artificially labeling the target samples can result in a good representation. Tri-training leverages three classifiers equally to provide pseudo-labels to unlabeled samples; however, the method does not assume labeling samples generated from a different domain. In this paper, we propose the use of an asymmetric tri-training method for unsupervised domain adaptation, where we assign pseudo-labels to unlabeled samples and train the neural networks as if they are true labels. In our work, we use three networks asymmetrically, and by asymmetric, we mean that two networks are used to label unlabeled target samples, and one network is trained by the pseudo-labeled samples to obtain target-discriminative representations. Our proposed method was shown to achieve a state-of-the-art performance on the benchmark digit recognition datasets for domain adaptation.", "", "", "In this paper, we propose a new unsupervised domain adaptation approach called Collaborative and Adversarial Network (CAN) through domain-collaborative and domain-adversarial training of neural networks. We add several domain classifiers on multiple CNN feature extraction blocks1, in which each domain classifier is connected to the hidden representations from one block and one loss function is defined based on the hidden presentation and the domain labels (e.g., source and target). We design a new loss function by integrating the losses from all blocks in order to learn domain informative representations from lower blocks through collaborative learning and learn domain uninformative representations from higher blocks through adversarial learning. We further extend our CAN method as Incremental CAN (iCAN), in which we iteratively select a set of pseudo-labelled target samples based on the image classifier and the last domain classifier from the previous training epoch and re-train our CAN model by using the enlarged training set. Comprehensive experiments on two benchmark datasets Office and ImageCLEF-DA clearly demonstrate the effectiveness of our newly proposed approaches CAN and iCAN for unsupervised domain adaptation.", "Transfer learning is established as an effective technology in computer vision for leveraging rich labeled data in the source domain to build an accurate classifier for the target domain. However, most prior methods have not simultaneously reduced the difference in both the marginal distribution and conditional distribution between domains. In this paper, we put forward a novel transfer learning approach, referred to as Joint Distribution Adaptation (JDA). Specifically, JDA aims to jointly adapt both the marginal distribution and conditional distribution in a principled dimensionality reduction procedure, and construct new feature representation that is effective and robust for substantial distribution difference. Extensive experiments verify that JDA can significantly outperform several state-of-the-art methods on four types of cross-domain image classification problems." ] }
1811.08599	2901518678	Most existing virtual try-on applications require clean clothes images. Instead, we present a novel virtual Try-On network, M2E-Try On Net, which transfers the clothes from a model image to a person image without the need of any clean product images. To obtain a realistic image of person wearing the desired model clothes, we aim to solve the following challenges: 1) non-rigid nature of clothes - we need to align poses between the model and the user; 2) richness in textures of fashion items - preserving the fine details and characteristics of the clothes is critical for photo-realistic transfer; 3) variation of identity appearances - it is required to fit the desired model clothes to the person identity seamlessly. To tackle these challenges, we introduce three key components, including the pose alignment network (PAN), the texture refinement network (TRN) and the fitting network (FTN). Since it is unlikely to gather image pairs of input person image and desired output image (i.e. person wearing the desired clothes), our framework is trained in a self-supervised manner to gradually transfer the poses and textures of the model's clothes to the desired appearance. In the experiments, we verify on the Deep Fashion dataset and MVC dataset that our method can generate photo-realistic images for the person to try-on the model clothes. Furthermore, we explore the model capability for different fashion items, including both upper and lower garments.	Human parsing has been studied for fine-grained segmentation of human body parts @cite_19 @cite_18 . @cite_25 extended object segmentation to object part-level segmentation and released the PASCAL PART dataset including pixel-level part annotations of the human body. @cite_23 collected a new person dataset (LIP) for human parsing and fashion clothes segmentation. @cite_27 proposed a multi-path refinement network to achieve high resolution and accurate part segmentation. Our work exploits @cite_23 to extract the region of interest that covers the clothes part in person images.	{ "cite_N": [ "@cite_18", "@cite_19", "@cite_27", "@cite_23", "@cite_25" ], "mid": [ "", "2890259823", "2563705555", "2598915960", "2104408738" ], "abstract": [ "", "Fully convolutional networks (FCN) have achieved great success in human parsing in recent years. In conventional human parsing tasks, pixel-level labeling is required for guiding the training, which usually involves enormous human labeling efforts. To ease the labeling efforts, we propose a novel weakly supervised human parsing method which only requires simple object keypoint annotations for learning. We develop an iterative learning method to generate pseudo part segmentation masks from keypoint labels. With these pseudo masks, we train an FCN network to output pixel-level human parsing predictions. Furthermore, we develop a correlation network to perform joint prediction of part and object segmentation masks and improve the segmentation performance. The experiment results show that our weakly supervised method is able to achieve very competitive human parsing results. Despite our method only uses simple keypoint annotations for learning, we are able to achieve comparable performance with fully supervised methods which use the expensive pixel-level annotations.", "Recently, very deep convolutional neural networks (CNNs) have shown outstanding performance in object recognition and have also been the first choice for dense classification problems such as semantic segmentation. However, repeated subsampling operations like pooling or convolution striding in deep CNNs lead to a significant decrease in the initial image resolution. Here, we present RefineNet, a generic multi-path refinement network that explicitly exploits all the information available along the down-sampling process to enable high-resolution prediction using long-range residual connections. In this way, the deeper layers that capture high-level semantic features can be directly refined using fine-grained features from earlier convolutions. The individual components of RefineNet employ residual connections following the identity mapping mindset, which allows for effective end-to-end training. Further, we introduce chained residual pooling, which captures rich background context in an efficient manner. We carry out comprehensive experiments and set new state-of-the-art results on seven public datasets. In particular, we achieve an intersection-over-union score of 83.4 on the challenging PASCAL VOC 2012 dataset, which is the best reported result to date.", "Human parsing has recently attracted a lot of research interests due to its huge application potentials. However existing datasets have limited number of images and annotations, and lack the variety of human appearances and the coverage of challenging cases in unconstrained environment. In this paper, we introduce a new benchmark Look into Person (LIP) that makes a significant advance in terms of scalability, diversity and difficulty, a contribution that we feel is crucial for future developments in human-centric analysis. This comprehensive dataset contains over 50,000 elaborately annotated images with 19 semantic part labels, which are captured from a wider range of viewpoints, occlusions and background complexity. Given these rich annotations we perform detailed analysis of the leading human parsing approaches, gaining insights into the success and failures of these methods. Furthermore, in contrast to the existing efforts on improving the feature discriminative capability, we solve human parsing by exploring a novel self-supervised structure-sensitive learning approach, which imposes human pose structures into parsing results without resorting to extra supervision (i.e., no need for specifically labeling human joints in model training). Our self-supervised learning framework can be injected into any advanced neural networks to help incorporate rich high-level knowledge regarding human joints from a global perspective and improve the parsing results. Extensive evaluations on our LIP and the public PASCAL-Person-Part dataset demonstrate the superiority of our method.", "Detecting objects becomes difficult when we need to deal with large shape deformation, occlusion and low resolution. We propose a novel approach to i) handle large deformations and partial occlusions in animals (as examples of highly deformable objects), ii) describe them in terms of body parts, and iii) detect them when their body parts are hard to detect (e.g., animals depicted at low resolution). We represent the holistic object and body parts separately and use a fully connected model to arrange templates for the holistic object and body parts. Our model automatically decouples the holistic object or body parts from the model when they are hard to detect. This enables us to represent a large number of holistic object and body part combinations to better deal with different \"detectability\" patterns caused by deformations, occlusion and or low resolution. We apply our method to the six animal categories in the PASCAL VOC dataset and show that our method significantly improves state-of-the-art (by 4.1 AP) and provides a richer representation for objects. During training we use annotations for body parts (e.g., head, torso, etc.), making use of a new dataset of fully annotated object parts for PASCAL VOC 2010, which provides a mask for each part." ] }
1811.08599	2901518678	Most existing virtual try-on applications require clean clothes images. Instead, we present a novel virtual Try-On network, M2E-Try On Net, which transfers the clothes from a model image to a person image without the need of any clean product images. To obtain a realistic image of person wearing the desired model clothes, we aim to solve the following challenges: 1) non-rigid nature of clothes - we need to align poses between the model and the user; 2) richness in textures of fashion items - preserving the fine details and characteristics of the clothes is critical for photo-realistic transfer; 3) variation of identity appearances - it is required to fit the desired model clothes to the person identity seamlessly. To tackle these challenges, we introduce three key components, including the pose alignment network (PAN), the texture refinement network (TRN) and the fitting network (FTN). Since it is unlikely to gather image pairs of input person image and desired output image (i.e. person wearing the desired clothes), our framework is trained in a self-supervised manner to gradually transfer the poses and textures of the model's clothes to the desired appearance. In the experiments, we verify on the Deep Fashion dataset and MVC dataset that our method can generate photo-realistic images for the person to try-on the model clothes. Furthermore, we explore the model capability for different fashion items, including both upper and lower garments.	Apart from human parsing for part segmentation, the works in @cite_15 @cite_21 studied human pose estimation for pose analysis. @cite_15 proposed a Part Affinity Fields for human pose estimation based on key points. Later, to achieve more accurate pose estimation, DensePose @cite_21 proposed dense human pose estimation method by mapping each pixel to a dense pose point. In our work, we utilize the estimated dense poses for clothes region warping and pose alignment.	{ "cite_N": [ "@cite_15", "@cite_21" ], "mid": [ "2559085405", "2963876278" ], "abstract": [ "We present an approach to efficiently detect the 2D pose of multiple people in an image. The approach uses a nonparametric representation, which we refer to as Part Affinity Fields (PAFs), to learn to associate body parts with individuals in the image. The architecture encodes global context, allowing a greedy bottom-up parsing step that maintains high accuracy while achieving realtime performance, irrespective of the number of people in the image. The architecture is designed to jointly learn part locations and their association via two branches of the same sequential prediction process. Our method placed first in the inaugural COCO 2016 keypoints challenge, and significantly exceeds the previous state-of-the-art result on the MPII Multi-Person benchmark, both in performance and efficiency.", "In this work we establish dense correspondences between an RGB image and a surface-based representation of the human body, a task we refer to as dense human pose estimation. We gather dense correspondences for 50K persons appearing in the COCO dataset by introducing an efficient annotation pipeline. We then use our dataset to train CNN-based systems that deliver dense correspondence 'in the wild', namely in the presence of background, occlusions and scale variations. We improve our training set's effectiveness by training an inpainting network that can fill in missing ground truth values and report improvements with respect to the best results that would be achievable in the past. We experiment with fully-convolutional networks and region-based models and observe a superiority of the latter. We further improve accuracy through cascading, obtaining a system that delivers highly-accurate results at multiple frames per second on a single gpu. Supplementary materials, data, code, and videos are provided on the project page http: densepose.org." ] }
1811.08599	2901518678	Most existing virtual try-on applications require clean clothes images. Instead, we present a novel virtual Try-On network, M2E-Try On Net, which transfers the clothes from a model image to a person image without the need of any clean product images. To obtain a realistic image of person wearing the desired model clothes, we aim to solve the following challenges: 1) non-rigid nature of clothes - we need to align poses between the model and the user; 2) richness in textures of fashion items - preserving the fine details and characteristics of the clothes is critical for photo-realistic transfer; 3) variation of identity appearances - it is required to fit the desired model clothes to the person identity seamlessly. To tackle these challenges, we introduce three key components, including the pose alignment network (PAN), the texture refinement network (TRN) and the fitting network (FTN). Since it is unlikely to gather image pairs of input person image and desired output image (i.e. person wearing the desired clothes), our framework is trained in a self-supervised manner to gradually transfer the poses and textures of the model's clothes to the desired appearance. In the experiments, we verify on the Deep Fashion dataset and MVC dataset that our method can generate photo-realistic images for the person to try-on the model clothes. Furthermore, we explore the model capability for different fashion items, including both upper and lower garments.	Generative adversarial network (GAN) @cite_10 has been used for image-based generation. Recently, GAN has been used for person image generation @cite_5 to generate the human image from pose representation. @cite_22 proposed a generative network to generate fashion images from textual inputs. For fashion image generation, a more intuitive way is to generate images from a person image and the desired clothes image, i.e. a virtual try-on system. This virtual try-on task has been studied in the past a few years @cite_24 @cite_8 @cite_6 @cite_3 . Inspired by cycle image translation @cite_24 , @cite_8 proposed to translate person clothes using product clothes as the condition. To preserve a detailed texture information, @cite_6 proposed to generate a synthesized image from clothes image and clothing-agnostic person representation. @cite_3 proposed to refine the clothes detail information by adding a warped product image by applying Geometric Matching Module (GMM). However, these methods rely on product images as the input. Instead, our method focuses on transferring the dressed clothes on an arbitrary model image without the need for clean product images.	{ "cite_N": [ "@cite_22", "@cite_8", "@cite_6", "@cite_3", "@cite_24", "@cite_5", "@cite_10" ], "mid": [ "2964318046", "2964094136", "2964050021", "2883309205", "2962793481", "", "2099471712" ], "abstract": [ "We present a novel and effective approach for generating new clothing on a wearer through generative adversarial learning. Given an input image of a person and a sentence describing a different outfit, our model “redresses” the person as desired, while at the same time keeping the wearer and her his pose unchanged. Generating new outfits with precise regions conforming to a language description while retaining wearer’s body structure is a new challenging task. Existing generative adversarial networks are not ideal in ensuring global coherence of structure given both the input photograph and language description as conditions. We address this challenge by decomposing the complex generative process into two conditional stages. In the first stage, we generate a plausible semantic segmentation map that obeys the wearer’s pose as a latent spatial arrangement. An effective spatial constraint is formulated to guide the generation of this semantic segmentation map. In the second stage, a generative model with a newly proposed compositional mapping layer is used to render the final image with precise regions and textures conditioned on this map. We extended the DeepFashion dataset [8] by collecting sentence descriptions for 79K images. We demonstrate the effectiveness of our approach through both quantitative and qualitative evaluations. A user study is also conducted.", "We present a novel method to solve image analogy problems [3]: it allows to learn the relation between paired images present in training data, and then generalize and generate images that correspond to the relation, but were never seen in the training set. Therefore, we call the method Conditional Analogy Generative Adversarial Network (CAGAN), as it is based on adversarial training and employs deep convolutional neural networks. An especially interesting application of that technique is automatic swapping of clothing on fashion model photos. Our work has the following contributions. First, the definition of the end-to-end trainable CAGAN architecture, which implicitly learns segmentation masks without expensive supervised labeling data. Second, experimental results show plausible segmentation masks and often convincing swapped images, given the target article. Finally, we discuss the next steps for that technique: neural network architecture improvements and more advanced applications.", "We present an image-based VIirtual Try-On Network (VITON) without using 3D information in any form, which seamlessly transfers a desired clothing item onto the corresponding region of a person using a coarse-to-fine strategy. Conditioned upon a new clothing-agnostic yet descriptive person representation, our framework first generates a coarse synthesized image with the target clothing item overlaid on that same person in the same pose. We further enhance the initial blurry clothing area with a refinement network. The network is trained to learn how much detail to utilize from the target clothing item, and where to apply to the person in order to synthesize a photo-realistic image in which the target item deforms naturally with clear visual patterns. Experiments on our newly collected Zalando dataset demonstrate its promise in the image-based virtual try-on task over state-of-the-art generative models.1", "Image-based virtual try-on systems for fitting a new in-shop clothes into a person image have attracted increasing research attention, yet is still challenging. A desirable pipeline should not only transform the target clothes into the most fitting shape seamlessly but also preserve well the clothes identity in the generated image, that is, the key characteristics (e.g. texture, logo, embroidery) that depict the original clothes. However, previous image-conditioned generation works fail to meet these critical requirements towards the plausible virtual try-on performance since they fail to handle large spatial misalignment between the input image and target clothes. Prior work explicitly tackled spatial deformation using shape context matching, but failed to preserve clothing details due to its coarse-to-fine strategy. In this work, we propose a new fully-learnable Characteristic-Preserving Virtual Try-On Network (CP-VTON) for addressing all real-world challenges in this task. First, CP-VTON learns a thin-plate spline transformation for transforming the in-shop clothes into fitting the body shape of the target person via a new Geometric Matching Module (GMM) rather than computing correspondences of interest points as prior works did. Second, to alleviate boundary artifacts of warped clothes and make the results more realistic, we employ a Try-On Module that learns a composition mask to integrate the warped clothes and the rendered image to ensure smoothness. Extensive experiments on a fashion dataset demonstrate our CP-VTON achieves the state-of-the-art virtual try-on performance both qualitatively and quantitatively.", "Image-to-image translation is a class of vision and graphics problems where the goal is to learn the mapping between an input image and an output image using a training set of aligned image pairs. However, for many tasks, paired training data will not be available. We present an approach for learning to translate an image from a source domain X to a target domain Y in the absence of paired examples. Our goal is to learn a mapping G : X → Y such that the distribution of images from G(X) is indistinguishable from the distribution Y using an adversarial loss. Because this mapping is highly under-constrained, we couple it with an inverse mapping F : Y → X and introduce a cycle consistency loss to push F(G(X)) ≈ X (and vice versa). Qualitative results are presented on several tasks where paired training data does not exist, including collection style transfer, object transfiguration, season transfer, photo enhancement, etc. Quantitative comparisons against several prior methods demonstrate the superiority of our approach.", "", "We propose a new framework for estimating generative models via an adversarial process, in which we simultaneously train two models: a generative model G that captures the data distribution, and a discriminative model D that estimates the probability that a sample came from the training data rather than G. The training procedure for G is to maximize the probability of D making a mistake. This framework corresponds to a minimax two-player game. In the space of arbitrary functions G and D, a unique solution exists, with G recovering the training data distribution and D equal to ½ everywhere. In the case where G and D are defined by multilayer perceptrons, the entire system can be trained with backpropagation. There is no need for any Markov chains or unrolled approximate inference networks during either training or generation of samples. Experiments demonstrate the potential of the framework through qualitative and quantitative evaluation of the generated samples." ] }
1811.08599	2901518678	Most existing virtual try-on applications require clean clothes images. Instead, we present a novel virtual Try-On network, M2E-Try On Net, which transfers the clothes from a model image to a person image without the need of any clean product images. To obtain a realistic image of person wearing the desired model clothes, we aim to solve the following challenges: 1) non-rigid nature of clothes - we need to align poses between the model and the user; 2) richness in textures of fashion items - preserving the fine details and characteristics of the clothes is critical for photo-realistic transfer; 3) variation of identity appearances - it is required to fit the desired model clothes to the person identity seamlessly. To tackle these challenges, we introduce three key components, including the pose alignment network (PAN), the texture refinement network (TRN) and the fitting network (FTN). Since it is unlikely to gather image pairs of input person image and desired output image (i.e. person wearing the desired clothes), our framework is trained in a self-supervised manner to gradually transfer the poses and textures of the model's clothes to the desired appearance. In the experiments, we verify on the Deep Fashion dataset and MVC dataset that our method can generate photo-realistic images for the person to try-on the model clothes. Furthermore, we explore the model capability for different fashion items, including both upper and lower garments.	Similarly, some recent works @cite_12 @cite_14 @cite_17 have been proposed to transfer the person images to different variations with the corresponding human poses. @cite_21 proposed a pose transfer network based on dense pose condition. The methods in @cite_17 and @cite_14 use human part segmentation to guide the human pose translation. These methods are merely based on the generative models which often fail to recover the textures of garments. In contrast, our method is able to preserve the characteristics of clothes. At the same time, our method can transfer the garment from the model person to the target person while preserving the identity and pose.	{ "cite_N": [ "@cite_14", "@cite_21", "@cite_12", "@cite_17" ], "mid": [ "2895668250", "2963876278", "2890816492", "2892217449" ], "abstract": [ "We present Swapnet, a framework to transfer garments across images of people with arbitrary body pose, shape, and clothing. Garment transfer is a challenging task that requires (i) disentangling the features of the clothing from the body pose and shape and (ii) realistic synthesis of the garment texture on the new body. We present a neural network architecture that tackles these sub-problems with two task-specific sub-networks. Since acquiring pairs of images showing the same clothing on different bodies is difficult, we propose a novel weakly-supervised approach that generates training pairs from a single image via data augmentation. We present the first fully automatic method for garment transfer in unconstrained images without solving the difficult 3D reconstruction problem. We demonstrate a variety of transfer results and highlight our advantages over traditional image-to-image and analogy pipelines.", "In this work we establish dense correspondences between an RGB image and a surface-based representation of the human body, a task we refer to as dense human pose estimation. We gather dense correspondences for 50K persons appearing in the COCO dataset by introducing an efficient annotation pipeline. We then use our dataset to train CNN-based systems that deliver dense correspondence 'in the wild', namely in the presence of background, occlusions and scale variations. We improve our training set's effectiveness by training an inpainting network that can fill in missing ground truth values and report improvements with respect to the best results that would be achievable in the past. We experiment with fully-convolutional networks and region-based models and observe a superiority of the latter. We further improve accuracy through cascading, obtaining a system that delivers highly-accurate results at multiple frames per second on a single gpu. Supplementary materials, data, code, and videos are provided on the project page http: densepose.org.", "In this work we integrate ideas from surface-based modeling with neural synthesis: we propose a combination of surface-based pose estimation and deep generative models that allows us to perform accurate pose transfer, i.e. synthesize a new image of a person based on a single image of that person and the image of a pose donor. We use a dense pose estimation system that maps pixels from both images to a common surface-based coordinate system, allowing the two images to be brought in correspondence with each other. We inpaint and refine the source image intensities in the surface coordinate system, prior to warping them onto the target pose. These predictions are fused with those of a convolutional predictive module through a neural synthesis module allowing for training the whole pipeline jointly end-to-end, optimizing a combination of adversarial and perceptual losses. We show that dense pose estimation is a substantially more powerful conditioning input than landmark-, or mask-based alternatives, and report systematic improvements over state of the art generators on DeepFashion and MVC datasets.", "Despite remarkable advances in image synthesis research, existing works often fail in manipulating images under the context of large geometric transformations. Synthesizing person images conditioned on arbitrary poses is one of the most representative examples where the generation quality largely relies on the capability of identifying and modeling arbitrary transformations on different body parts. Current generative models are often built on local convolutions and overlook the key challenges (e.g. heavy occlusions, different views or dramatic appearance changes) when distinct geometric changes happen for each part, caused by arbitrary pose manipulations. This paper aims to resolve these challenges induced by geometric variability and spatial displacements via a new Soft-Gated Warping Generative Adversarial Network (Warping-GAN), which is composed of two stages: 1) it first synthesizes a target part segmentation map given a target pose, which depicts the region-level spatial layouts for guiding image synthesis with higher-level structure constraints; 2) the Warping-GAN equipped with a soft-gated warping-block learns feature-level mapping to render textures from the original image into the generated segmentation map. Warping-GAN is capable of controlling different transformation degrees given distinct target poses. Moreover, the proposed warping-block is light-weight and flexible enough to be injected into any networks. Human perceptual studies and quantitative evaluations demonstrate the superiority of our Warping-GAN that significantly outperforms all existing methods on two large datasets." ] }
1811.08599	2901518678	Most existing virtual try-on applications require clean clothes images. Instead, we present a novel virtual Try-On network, M2E-Try On Net, which transfers the clothes from a model image to a person image without the need of any clean product images. To obtain a realistic image of person wearing the desired model clothes, we aim to solve the following challenges: 1) non-rigid nature of clothes - we need to align poses between the model and the user; 2) richness in textures of fashion items - preserving the fine details and characteristics of the clothes is critical for photo-realistic transfer; 3) variation of identity appearances - it is required to fit the desired model clothes to the person identity seamlessly. To tackle these challenges, we introduce three key components, including the pose alignment network (PAN), the texture refinement network (TRN) and the fitting network (FTN). Since it is unlikely to gather image pairs of input person image and desired output image (i.e. person wearing the desired clothes), our framework is trained in a self-supervised manner to gradually transfer the poses and textures of the model's clothes to the desired appearance. In the experiments, we verify on the Deep Fashion dataset and MVC dataset that our method can generate photo-realistic images for the person to try-on the model clothes. Furthermore, we explore the model capability for different fashion items, including both upper and lower garments.	Our work uses several fashion datasets @cite_11 @cite_2 for training and evaluating our try-on network. Deep Fashion dataset @cite_11 is a fashion dataset for clothes attribute prediction and landmark detection. MVC dataset @cite_2 is for view-invariant clothing retrieval and attribute prediction. These datasets provide us a large number of dressed person images with multiple views, poses and identities.	{ "cite_N": [ "@cite_2", "@cite_11" ], "mid": [ "2410358280", "2471768434" ], "abstract": [ "Clothing retrieval and clothing style recognition are important and practical problems. They have drawn a lot of attention in recent years. However, the clothing photos collected in existing datasets are mostly of front- or near-front view. There are no datasets designed to study the influences of different viewing angles on clothing retrieval performance. To address view-invariant clothing retrieval problem properly, we construct a challenge clothing dataset, called Multi-View Clothing dataset. This dataset not only has four different views for each clothing item, but also provides 264 attributes for describing clothing appearance. We adopt a state-of-the-art deep learning method to present baseline results for the attribute prediction and clothing retrieval performance. We also evaluate the method on a more difficult setting, cross-view exact clothing item retrieval. Our dataset will be made publicly available for further studies towards view-invariant clothing retrieval.", "Recent advances in clothes recognition have been driven by the construction of clothes datasets. Existing datasets are limited in the amount of annotations and are difficult to cope with the various challenges in real-world applications. In this work, we introduce DeepFashion1, a large-scale clothes dataset with comprehensive annotations. It contains over 800,000 images, which are richly annotated with massive attributes, clothing landmarks, and correspondence of images taken under different scenarios including store, street snapshot, and consumer. Such rich annotations enable the development of powerful algorithms in clothes recognition and facilitating future researches. To demonstrate the advantages of DeepFashion, we propose a new deep model, namely FashionNet, which learns clothing features by jointly predicting clothing attributes and landmarks. The estimated landmarks are then employed to pool or gate the learned features. It is optimized in an iterative manner. Extensive experiments demonstrate the effectiveness of FashionNet and the usefulness of DeepFashion." ] }
1811.08853	2901888743	In discussions hosted on discussion forums for MOOCs, references to online learning resources are often of central importance. They contextualize the discussion, anchoring the discussion participants' presentation of the issues and their understanding. However they are usually mentioned in free text, without appropriate hyperlinking to their associated resource. Automated learning resource mention hyperlinking and categorization will facilitate discussion and searching within MOOC forums, and also benefit the contextualization of such resources across disparate views. We propose the novel problem of learning resource mention identification in MOOC forums. As this is a novel task with no publicly available data, we first contribute a large-scale labeled dataset, dubbed the Forum Resource Mention (FoRM) dataset, to facilitate our current research and future research on this task. We then formulate this task as a sequence tagging problem and investigate solution architectures to address the problem. Importantly, we identify two major challenges that hinder the application of sequence tagging models to the task: (1) the diversity of resource mention expression, and (2) long-range contextual dependencies. We address these challenges by incorporating character-level and thread context information into a LSTM-CRF model. First, we incorporate a character encoder to address the out-of-vocabulary problem caused by the diversity of mention expressions. Second, to address the context dependency challenge, we encode thread contexts using an RNN-based context encoder, and apply the attention mechanism to selectively leverage useful context information during sequence tagging. Experiments on FoRM show that the proposed method improves the baseline deep sequence tagging models notably, significantly bettering performance on instances that exemplify the two challenges.	One problem of LSTM--CRF is that it only captures the word-level semantics. This causes a problem when intra-word morphological and character-level information are also very important for recognizing named entities. Recently, @cite_16 augmented the work of @cite_32 with character-level CNNs. Chiu and Nichols @cite_35 incorporated the character-level CNN to BLSTM and achieved a better performance in NER. In our task, resource mention identification, the widely existing OOV words, such as Q1'', Q2'', hw2'' in Figure , greatly increase the difficulty of capturing word-semantics. Therefore, we also incorporate the character-level semantics by proposing a character encoder via LSTM.	{ "cite_N": [ "@cite_35", "@cite_16", "@cite_32" ], "mid": [ "2963625095", "1951325712", "2158899491" ], "abstract": [ "Named entity recognition is a challenging task that has traditionally required large amounts of knowledge in the form of feature engineering and lexicons to achieve high performance. In this paper, we present a novel neural network architecture that automatically detects word- and character-level features using a hybrid bidirectional LSTM and CNN architecture, eliminating the need for most feature engineering. We also propose a novel method of encoding partial lexicon matches in neural networks and compare it to existing approaches. Extensive evaluation shows that, given only tokenized text and publicly available word embeddings, our system is competitive on the CoNLL-2003 dataset and surpasses the previously reported state of the art performance on the OntoNotes 5.0 dataset by 2.13 F1 points. By using two lexicons constructed from publicly-available sources, we establish new state of the art performance with an F1 score of 91.62 on CoNLL-2003 and 86.28 on OntoNotes, surpassing systems that employ heavy feature engineering, proprietary lexicons, and rich entity linking information.", "Most state-of-the-art named entity recognition (NER) systems rely on handcrafted features and on the output of other NLP tasks such as part-of-speech (POS) tagging and text chunking. In this work we propose a language-independent NER system that uses automatically learned features only. Our approach is based on the CharWNN deep neural network, which uses word-level and character-level representations (embeddings) to perform sequential classification. We perform an extensive number of experiments using two annotated corpora in two different languages: HAREM I corpus, which contains texts in Portuguese; and the SPA CoNLL-2002 corpus, which contains texts in Spanish. Our experimental results shade light on the contribution of neural character embeddings for NER. Moreover, we demonstrate that the same neural network which has been successfully applied to POS tagging can also achieve state-of-the-art results for language-independet NER, using the same hyperparameters, and without any handcrafted features. For the HAREM I corpus, CharWNN outperforms the state-of-the-art system by 7.9 points in the F1-score for the total scenario (ten NE classes), and by 7.2 points in the F1 for the selective scenario (five NE classes).", "We propose a unified neural network architecture and learning algorithm that can be applied to various natural language processing tasks including part-of-speech tagging, chunking, named entity recognition, and semantic role labeling. This versatility is achieved by trying to avoid task-specific engineering and therefore disregarding a lot of prior knowledge. Instead of exploiting man-made input features carefully optimized for each task, our system learns internal representations on the basis of vast amounts of mostly unlabeled training data. This work is then used as a basis for building a freely available tagging system with good performance and minimal computational requirements." ] }
1811.08853	2901888743	In discussions hosted on discussion forums for MOOCs, references to online learning resources are often of central importance. They contextualize the discussion, anchoring the discussion participants' presentation of the issues and their understanding. However they are usually mentioned in free text, without appropriate hyperlinking to their associated resource. Automated learning resource mention hyperlinking and categorization will facilitate discussion and searching within MOOC forums, and also benefit the contextualization of such resources across disparate views. We propose the novel problem of learning resource mention identification in MOOC forums. As this is a novel task with no publicly available data, we first contribute a large-scale labeled dataset, dubbed the Forum Resource Mention (FoRM) dataset, to facilitate our current research and future research on this task. We then formulate this task as a sequence tagging problem and investigate solution architectures to address the problem. Importantly, we identify two major challenges that hinder the application of sequence tagging models to the task: (1) the diversity of resource mention expression, and (2) long-range contextual dependencies. We address these challenges by incorporating character-level and thread context information into a LSTM-CRF model. First, we incorporate a character encoder to address the out-of-vocabulary problem caused by the diversity of mention expressions. Second, to address the context dependency challenge, we encode thread contexts using an RNN-based context encoder, and apply the attention mechanism to selectively leverage useful context information during sequence tagging. Experiments on FoRM show that the proposed method improves the baseline deep sequence tagging models notably, significantly bettering performance on instances that exemplify the two challenges.	In computational linguistics, anaphora is typically defined as references to items mentioned earlier in the discourse or pointing back'' reference as described by @cite_36 . is then defined as resolving anaphora to its corresponding entities in a discourse. Resolving repeated references to an entity is similar to differentiating whether a mention is a valid resource mention within the course.	{ "cite_N": [ "@cite_36" ], "mid": [ "1657765689" ], "abstract": [ "This paper presents a multilingual robust, knowledge-poor approach to resolving pronouns in technical manuals. This approach is a modification of the practical approach (Mitkov 1998a) and operates on texts pre-processed by a part-of-speech tagger. Input is checked against agreement and a number of antecedent indicators. Candidates are assigned scores by each indicator and the candidate with the highest aggregate score is returned as the antecedent. We propose this approach as a platform for multilingual pronoun resolution. The robust approach was initially developed and tested for English, but we have also adapted and tested it for Polish and Arabic. For both languages, we found that adaptation required minimum modification and that further, even if used unmodified, the approach delivers acceptable success rates. Preliminary evaluation reports high success rates in the range of over 90 ." ] }
1811.08622	2901420404	How to obtain the desirable representation of a 3D shape, which is discriminative across categories and polymerized within classes, is a significant challenge in 3D shape retrieval. Most existing 3D shape retrieval methods focus on capturing strong discriminative shape representation with softmax loss for the classification task, while the shape feature learning with metric loss is neglected for 3D shape retrieval. In this paper, we address this problem based on the intuition that the cosine distance of shape embeddings should be close enough within the same class and far away across categories. Since most of 3D shape retrieval tasks use cosine distance of shape features for measuring shape similarity, we propose a novel metric loss named angular triplet-center loss, which directly optimizes the cosine distances between the features. It inherits the triplet-center loss property to achieve larger inter-class distance and smaller intra-class distance simultaneously. Unlike previous metric loss utilized in 3D shape retrieval methods, where Euclidean distance is adopted and the margin design is difficult, the proposed method is more convenient to train feature embeddings and more suitable for 3D shape retrieval. Moreover, the angle margin is adopted to replace the cosine margin in order to provide more explicit discriminative constraints on an embedding space. Extensive experimental results on two popular 3D object retrieval benchmarks, ModelNet40 and ShapeNetCore 55, demonstrate the effectiveness of our proposed loss, and our method has achieved state-of-the-art results on various 3D shape datasets.	A large number of works @cite_28 have been proposed to address 3D shape retrieval problem in recent years. In particular, with the explosive growth of large-scale public 3D shape repositories and the success made by the convolutional neural network (CNN) in computer vision, the CNN-based 3D shape retrieval methods have achieved impressive performance. In this section, we will mainly introduce representative 3D shape retrieval methods based on deep learning techniques.	{ "cite_N": [ "@cite_28" ], "mid": [ "2604249033" ], "abstract": [ "Deep learning has recently gained popularity achieving state-of-the-art performance in tasks involving text, sound, or image processing. Due to its outstanding performance, there have been efforts to apply it in more challenging scenarios, for example, 3D data processing. This article surveys methods applying deep learning on 3D data and provides a classification based on how they exploit them. From the results of the examined works, we conclude that systems employing 2D views of 3D data typically surpass voxel-based (3D) deep models, which however, can perform better with more layers and severe data augmentation. Therefore, larger-scale datasets and increased resolutions are required." ] }
1811.08622	2901420404	How to obtain the desirable representation of a 3D shape, which is discriminative across categories and polymerized within classes, is a significant challenge in 3D shape retrieval. Most existing 3D shape retrieval methods focus on capturing strong discriminative shape representation with softmax loss for the classification task, while the shape feature learning with metric loss is neglected for 3D shape retrieval. In this paper, we address this problem based on the intuition that the cosine distance of shape embeddings should be close enough within the same class and far away across categories. Since most of 3D shape retrieval tasks use cosine distance of shape features for measuring shape similarity, we propose a novel metric loss named angular triplet-center loss, which directly optimizes the cosine distances between the features. It inherits the triplet-center loss property to achieve larger inter-class distance and smaller intra-class distance simultaneously. Unlike previous metric loss utilized in 3D shape retrieval methods, where Euclidean distance is adopted and the margin design is difficult, the proposed method is more convenient to train feature embeddings and more suitable for 3D shape retrieval. Moreover, the angle margin is adopted to replace the cosine margin in order to provide more explicit discriminative constraints on an embedding space. Extensive experimental results on two popular 3D object retrieval benchmarks, ModelNet40 and ShapeNetCore 55, demonstrate the effectiveness of our proposed loss, and our method has achieved state-of-the-art results on various 3D shape datasets.	Generally, 3D shape retrieval methods could be coarsely divided into two categories: model-based methods and view-based methods. The model-based method directly extracts the shape feature from the raw 3D representations of objects, such as polygon meshed @cite_4 @cite_24 , voxel grid @cite_27 @cite_5 , point cloud @cite_2 @cite_25 , or octree representation @cite_12 . For example, @cite_13 introduce two volumetric CNN network architectures including the auxiliary part-based classification task and the long anisotropic kernel for long-distance interactions, which improves the performance of volumetric CNN-based 3D shape recognition methods. @cite_0 propose kernel correlation and graph pooling to capture local patterns, leading a robust improvement for PointNet-like methods. @cite_12 performs 3D CNN on the octree representation of 3D shapes, showing more compact storage and fast computation than existing model-based approaches. Although model-based methods can effectively interpret the geometric characteristics of 3D shapes, their performances are limited by the high computational complexity (, voxel representation) and noise of naive 3D shape representation, such as incompleteness or occlusions.	{ "cite_N": [ "@cite_4", "@cite_24", "@cite_0", "@cite_27", "@cite_2", "@cite_5", "@cite_13", "@cite_25", "@cite_12" ], "mid": [ "", "2404723690", "2796040722", "2812373594", "2560609797", "2394951287", "", "2624503621", "2952789225" ], "abstract": [ "", "Establishing correspondence between shapes is a fundamental problem in geometry processing, arising in a wide variety of applications. The problem is especially difficult in the setting of non-isometric deformations, as well as in the presence of topological noise and missing parts, mainly due to the limited capability to model such deformations axiomatically. Several recent works showed that invariance to complex shape transformations can be learned from examples. In this paper, we introduce an intrinsic convolutional neural network architecture based on anisotropic diffusion kernels, which we term Anisotropic Convolutional Neural Network (ACNN). In our construction, we generalize convolutions to non-Euclidean domains by constructing a set of oriented anisotropic diffusion kernels, creating in this way a local intrinsic polar representation of the data ( patch'), which is then correlated with a filter. Several cascades of such filters, linear, and non-linear operators are stacked to form a deep neural network whose parameters are learned by minimizing a task-specific cost. We use ACNNs to effectively learn intrinsic dense correspondences between deformable shapes in very challenging settings, achieving state-of-the-art results on some of the most difficult recent correspondence benchmarks.", "Unlike on images, semantic learning on 3D point clouds using a deep network is challenging due to the naturally unordered data structure. Among existing works, PointNet has achieved promising results by directly learning on point sets. However, it does not take full advantage of a point's local neighborhood that contains fine-grained structural information which turns out to be helpful towards better semantic learning. In this regard, we present two new operations to improve PointNet with a more efficient exploitation of local structures. The first one focuses on local 3D geometric structures. In analogy to a convolution kernel for images, we define a point-set kernel as a set of learnable 3D points that jointly respond to a set of neighboring data points according to their geometric affinities measured by kernel correlation, adapted from a similar technique for point cloud registration. The second one exploits local high-dimensional feature structures by recursive feature aggregation on a nearest-neighbor-graph computed from 3D positions. Experiments show that our network can efficiently capture local information and robustly achieve better performances on major datasets. Our code is available at this http URL", "To address the high computational and memory cost in 3-D volumetric convolutional neural networks (CNNs), we propose an approach to train binary volumetric CNNs for 3-D object recognition. Our method is specifically designed for 3-D data, in which it transforms the inputs and weights in convolutional fully connected layers to binary values, which can potentially accelerate the networks by efficient bitwise operations. Two loss calculation methods are designed to solve the accuracy decrease problem when the weights in the last layer are binarized. Four binary volumetric CNNs are obtained from their corresponding floating-point networks using our approach. Evaluations on three public datasets from different domains (Computer Aided Design (CAD), light detection and ranging (LiDAR), and RGB-D) show that our binary volumetric CNNs can achieve a comparable recognition performance as their floating-point counterparts but consume less computational and memory resources.", "Point cloud is an important type of geometric data structure. Due to its irregular format, most researchers transform such data to regular 3D voxel grids or collections of images. This, however, renders data unnecessarily voluminous and causes issues. In this paper, we design a novel type of neural network that directly consumes point clouds, which well respects the permutation invariance of points in the input. Our network, named PointNet, provides a unified architecture for applications ranging from object classification, part segmentation, to scene semantic parsing. Though simple, PointNet is highly efficient and effective. Empirically, it shows strong performance on par or even better than state of the art. Theoretically, we provide analysis towards understanding of what the network has learnt and why the network is robust with respect to input perturbation and corruption.", "Building discriminative representations for 3D data has been an important task in computer graphics and computer vision research. Convolutional Neural Networks (CNNs) have shown to operate on 2D images with great success for a variety of tasks. Lifting convolution operators to 3D (3DCNNs) seems like a plausible and promising next step. Unfortunately, the computational complexity of 3D CNNs grows cubically with respect to voxel resolution. Moreover, since most 3D geometry representations are boundary based, occupied regions do not increase proportionately with the size of the discretization, resulting in wasted computation. In this work, we represent 3D spaces as volumetric fields, and propose a novel design that employs field probing filters to efficiently extract features from them. Each field probing filter is a set of probing points --- sensors that perceive the space. Our learning algorithm optimizes not only the weights associated with the probing points, but also their locations, which deforms the shape of the probing filters and adaptively distributes them in 3D space. The optimized probing points sense the 3D space \"intelligently\", rather than operating blindly over the entire domain. We show that field probing is significantly more efficient than 3DCNNs, while providing state-of-the-art performance, on classification tasks for 3D object recognition benchmark datasets.", "", "Few prior works study deep learning on point sets. PointNet by is a pioneer in this direction. However, by design PointNet does not capture local structures induced by the metric space points live in, limiting its ability to recognize fine-grained patterns and generalizability to complex scenes. In this work, we introduce a hierarchical neural network that applies PointNet recursively on a nested partitioning of the input point set. By exploiting metric space distances, our network is able to learn local features with increasing contextual scales. With further observation that point sets are usually sampled with varying densities, which results in greatly decreased performance for networks trained on uniform densities, we propose novel set learning layers to adaptively combine features from multiple scales. Experiments show that our network called PointNet++ is able to learn deep point set features efficiently and robustly. In particular, results significantly better than state-of-the-art have been obtained on challenging benchmarks of 3D point clouds.", "A longstanding question in computer vision concerns the representation of 3D shapes for recognition: should 3D shapes be represented with descriptors operating on their native 3D formats, such as voxel grid or polygon mesh, or can they be effectively represented with view-based descriptors? We address this question in the context of learning to recognize 3D shapes from a collection of their rendered views on 2D images. We first present a standard CNN architecture trained to recognize the shapes' rendered views independently of each other, and show that a 3D shape can be recognized even from a single view at an accuracy far higher than using state-of-the-art 3D shape descriptors. Recognition rates further increase when multiple views of the shapes are provided. In addition, we present a novel CNN architecture that combines information from multiple views of a 3D shape into a single and compact shape descriptor offering even better recognition performance. The same architecture can be applied to accurately recognize human hand-drawn sketches of shapes. We conclude that a collection of 2D views can be highly informative for 3D shape recognition and is amenable to emerging CNN architectures and their derivatives." ] }
1811.08622	2901420404	How to obtain the desirable representation of a 3D shape, which is discriminative across categories and polymerized within classes, is a significant challenge in 3D shape retrieval. Most existing 3D shape retrieval methods focus on capturing strong discriminative shape representation with softmax loss for the classification task, while the shape feature learning with metric loss is neglected for 3D shape retrieval. In this paper, we address this problem based on the intuition that the cosine distance of shape embeddings should be close enough within the same class and far away across categories. Since most of 3D shape retrieval tasks use cosine distance of shape features for measuring shape similarity, we propose a novel metric loss named angular triplet-center loss, which directly optimizes the cosine distances between the features. It inherits the triplet-center loss property to achieve larger inter-class distance and smaller intra-class distance simultaneously. Unlike previous metric loss utilized in 3D shape retrieval methods, where Euclidean distance is adopted and the margin design is difficult, the proposed method is more convenient to train feature embeddings and more suitable for 3D shape retrieval. Moreover, the angle margin is adopted to replace the cosine margin in order to provide more explicit discriminative constraints on an embedding space. Extensive experimental results on two popular 3D object retrieval benchmarks, ModelNet40 and ShapeNetCore 55, demonstrate the effectiveness of our proposed loss, and our method has achieved state-of-the-art results on various 3D shape datasets.	The view-based 3D shape retrieval methods represent a 3D shape with a collection of 2D projections. A typical deep learning example of the view-based technique is GIFT @cite_10 , extracting each single view feature by using CNN with GPU acceleration and adopting the inverted file to reduce computation in distance metrics. Apart from a single view, many researchers focus on building a compact shape feature through an informative multi-view image sequence of the 3D shape. For example, @cite_1 use a set of CNN to extract each view's feature and then aggregate information from multiple views into a single and compact shape representation with the element-wise maximum operation. @cite_11 develop a local MVCNN shape descriptors, which generates a local descriptor for any point on the shape and can be directly applicable to a wide range of shape analysis tasks. Recently, @cite_21 propose a view-group-shape framework (GVCNN) to exploit intrinsic hierarchical correlation and discriminability among views, achieving a significant performance gain on the 3D shape retrieval task. Typically, the deep representation of multiple views performs more discriminative for 3D shapes and leads the best retrieval results on various 3D shape datasets.	{ "cite_N": [ "@cite_1", "@cite_21", "@cite_10", "@cite_11" ], "mid": [ "", "2799162093", "2952195138", "2753872511" ], "abstract": [ "", "3D shape recognition has attracted much attention recently. Its recent advances advocate the usage of deep features and achieve the state-of-the-art performance. However, existing deep features for 3D shape recognition are restricted to a view-to-shape setting, which learns the shape descriptor from the view-level feature directly. Despite the exciting progress on view-based 3D shape description, the intrinsic hierarchical correlation and discriminability among views have not been well exploited, which is important for 3D shape representation. To tackle this issue, in this paper, we propose a group-view convolutional neural network (GVCNN) framework for hierarchical correlation modeling towards discriminative 3D shape description. The proposed GVCNN framework is composed of a hierarchical view-group-shape architecture, i.e., from the view level, the group level and the shape level, which are organized using a grouping strategy. Concretely, we first use an expanded CNN to extract a view level descriptor. Then, a grouping module is introduced to estimate the content discrimination of each view, based on which all views can be splitted into different groups according to their discriminative level. A group level description can be further generated by pooling from view descriptors. Finally, all group level descriptors are combined into the shape level descriptor according to their discriminative weights. Experimental results and comparison with state-of-the-art methods show that our proposed GVCNN method can achieve a significant performance gain on both the 3D shape classification and retrieval tasks.", "Projective analysis is an important solution for 3D shape retrieval, since human visual perceptions of 3D shapes rely on various 2D observations from different view points. Although multiple informative and discriminative views are utilized, most projection-based retrieval systems suffer from heavy computational cost, thus cannot satisfy the basic requirement of scalability for search engines. In this paper, we present a real-time 3D shape search engine based on the projective images of 3D shapes. The real-time property of our search engine results from the following aspects: (1) efficient projection and view feature extraction using GPU acceleration; (2) the first inverted file, referred as F-IF, is utilized to speed up the procedure of multi-view matching; (3) the second inverted file (S-IF), which captures a local distribution of 3D shapes in the feature manifold, is adopted for efficient context-based reranking. As a result, for each query the retrieval task can be finished within one second despite the necessary cost of IO overhead. We name the proposed 3D shape search engine, which combines GPU acceleration and Inverted File (Twice), as GIFT. Besides its high efficiency, GIFT also outperforms the state-of-the-art methods significantly in retrieval accuracy on various shape benchmarks and competitions.", "We present a new local descriptor for 3D shapes, directly applicable to a wide range of shape analysis problems such as point correspondences, semantic segmentation, affordance prediction, and shape-to-scan matching. The descriptor is produced by a convolutional network that is trained to embed geometrically and semantically similar points close to one another in descriptor space. The network processes surface neighborhoods around points on a shape that are captured at multiple scales by a succession of progressively zoomed-out views, taken from carefully selected camera positions. We leverage two extremely large sources of data to train our network. First, since our network processes rendered views in the form of 2D images, we repurpose architectures pretrained on massive image datasets. Second, we automatically generate a synthetic dense point correspondence dataset by nonrigid alignment of corresponding shape parts in a large collection of segmented 3D models. As a result of these design choices, our network effectively encodes multiscale local context and fine-grained surface detail. Our network can be trained to produce either category-specific descriptors or more generic descriptors by learning from multiple shape categories. Once trained, at test time, the network extracts local descriptors for shapes without requiring any part segmentation as input. Our method can produce effective local descriptors even for shapes whose category is unknown or different from the ones used while training. We demonstrate through several experiments that our learned local descriptors are more discriminative compared to state-of-the-art alternatives and are effective in a variety of shape analysis applications." ] }
1811.08824	2900783438	Increasingly, perceptual systems are being codified as strict pipelines wherein vision is treated as a pre-processing step to provide a dense representation of the scene to planners for high level reasoning downstream. Problematically, this paradigm forces models to represent nearly every aspect of the scene even if it has no bearing on the task at hand. In this work, we flip this paradigm, by introducing vision models whose feature representations are conditioned on embedded representations of the agent's goal. This allows the model to build scene descriptions that are specifically designed to help achieve that goal. We find this leads to models that learn faster, are substantially more parameter efficient and more robust than existing attention mechanisms in our domain. Our experiments are performed on a simulated robot item retrieval problem and trained in a fully end-to-end manner via imitation learning.	Learning to recognize objects, perform actions, and ground instructions to observed phenomena is core to many AI domains and advances are spread across the Robotics, Vision and Natural Language communities. Most immediately relevant to our work, is the recent proliferation of goal directed visual learning in simulated worlds @cite_29 @cite_41 @cite_43 @cite_47 @cite_1 @cite_12 which each aim to bring different amounts of language, vision and interaction to the task of navigating a 3D environment. These systems are often built using one of several open environment simulators based on 3D game engines @cite_4 @cite_38 @cite_30 @cite_16 . This has also been attempted in real 3D environments @cite_34 . Importantly, in contrast to our work, these approaches often pretrain as much of their networks as possible. @cite_43 do not pretrain for their RL based language learning. Their work focuses on a limited vocabulary with composition and does not address learning with occulusion or larger vocabularies.	{ "cite_N": [ "@cite_30", "@cite_38", "@cite_4", "@cite_41", "@cite_29", "@cite_1", "@cite_34", "@cite_43", "@cite_47", "@cite_16", "@cite_12" ], "mid": [ "", "2962887844", "", "", "2962684798", "", "2951660448", "2627585944", "2963800628", "", "" ], "abstract": [ "", "Two less addressed issues of deep reinforcement learning are (1) lack of generalization capability to new goals, and (2) data inefficiency, i.e., the model requires several (and often costly) episodes of trial and error to converge, which makes it impractical to be applied to real-world scenarios. In this paper, we address these two issues and apply our model to target-driven visual navigation. To address the first issue, we propose an actor-critic model whose policy is a function of the goal as well as the current state, which allows better generalization. To address the second issue, we propose the AI2-THOR framework, which provides an environment with high-quality 3D scenes and a physics engine. Our framework enables agents to take actions and interact with objects. Hence, we can collect a huge number of training samples efficiently. We show that our proposed method (1) converges faster than the state-of-the-art deep reinforcement learning methods, (2) generalizes across targets and scenes, (3) generalizes to a real robot scenario with a small amount of fine-tuning (although the model is trained in simulation), (4) is end-to-end trainable and does not need feature engineering, feature matching between frames or 3D reconstruction of the environment.", "", "", "We introduce Interactive Question Answering (IQA), the task of answering questions that require an autonomous agent to interact with a dynamic visual environment. IQA presents the agent with a scene and a question, like: \"Are there any apples in the fridge?\" The agent must navigate around the scene, acquire visual understanding of scene elements, interact with objects (e.g. open refrigerators) and plan for a series of actions conditioned on the question. Popular reinforcement learning approaches with a single controller perform poorly on IQA owing to the large and diverse state space. We propose the Hierarchical Interactive Memory Network (HIMN), consisting of a factorized set of controllers, allowing the system to operate at multiple levels of temporal abstraction. To evaluate HIMN, we introduce IQUAD V1, a new dataset built upon AI2-THOR [35], a simulated photo-realistic environment of configurable indoor scenes with interactive objects. IQUAD V1 has 75,000 questions, each paired with a unique scene configuration. Our experiments show that our proposed model outperforms popular single controller based methods on IQUAD V1. For sample questions and results, please view our video: https: youtu.be pXd3C-1jr98.", "", "We introduce a neural architecture for navigation in novel environments. Our proposed architecture learns to map from first-person views and plans a sequence of actions towards goals in the environment. The Cognitive Mapper and Planner (CMP) is based on two key ideas: a) a unified joint architecture for mapping and planning, such that the mapping is driven by the needs of the task, and b) a spatial memory with the ability to plan given an incomplete set of observations about the world. CMP constructs a top-down belief map of the world and applies a differentiable neural net planner to produce the next action at each time step. The accumulated belief of the world enables the agent to track visited regions of the environment. We train and test CMP on navigation problems in simulation environments derived from scans of real world buildings. Our experiments demonstrate that CMP outperforms alternate learning-based architectures, as well as, classical mapping and path planning approaches in many cases. Furthermore, it naturally extends to semantically specified goals, such as 'going to a chair'. We also deploy CMP on physical robots in indoor environments, where it achieves reasonable performance, even though it is trained entirely in simulation.", "We are increasingly surrounded by artificially intelligent technology that takes decisions and executes actions on our behalf. This creates a pressing need for general means to communicate with, instruct and guide artificial agents, with human language the most compelling means for such communication. To achieve this in a scalable fashion, agents must be able to relate language to the world and to actions; that is, their understanding of language must be grounded and embodied. However, learning grounded language is a notoriously challenging problem in artificial intelligence research. Here we present an agent that learns to interpret language in a simulated 3D environment where it is rewarded for the successful execution of written instructions. Trained via a combination of reinforcement and unsupervised learning, and beginning with minimal prior knowledge, the agent learns to relate linguistic symbols to emergent perceptual representations of its physical surroundings and to pertinent sequences of actions. The agent's comprehension of language extends beyond its prior experience, enabling it to apply familiar language to unfamiliar situations and to interpret entirely novel instructions. Moreover, the speed with which this agent learns new words increases as its semantic knowledge grows. This facility for generalising and bootstrapping semantic knowledge indicates the potential of the present approach for reconciling ambiguous natural language with the complexity of the physical world.", "A robot that can carry out a natural-language instruction has been a dream since before the Jetsons cartoon series imagined a life of leisure mediated by a fleet of attentive robot helpers. It is a dream that remains stubbornly distant. However, recent advances in vision and language methods have made incredible progress in closely related areas. This is significant because a robot interpreting a natural-language navigation instruction on the basis of what it sees is carrying out a vision and language process that is similar to Visual Question Answering. Both tasks can be interpreted as visually grounded sequence-to-sequence translation problems, and many of the same methods are applicable. To enable and encourage the application of vision and language methods to the problem of interpreting visually-grounded navigation instructions, we present the Matter-port3D Simulator - a large-scale reinforcement learning environment based on real imagery [11]. Using this simulator, which can in future support a range of embodied vision and language tasks, we provide the first benchmark dataset for visually-grounded natural language navigation in real buildings - the Room-to-Room (R2R) dataset1.", "", "" ] }
1811.08824	2900783438	Increasingly, perceptual systems are being codified as strict pipelines wherein vision is treated as a pre-processing step to provide a dense representation of the scene to planners for high level reasoning downstream. Problematically, this paradigm forces models to represent nearly every aspect of the scene even if it has no bearing on the task at hand. In this work, we flip this paradigm, by introducing vision models whose feature representations are conditioned on embedded representations of the agent's goal. This allows the model to build scene descriptions that are specifically designed to help achieve that goal. We find this leads to models that learn faster, are substantially more parameter efficient and more robust than existing attention mechanisms in our domain. Our experiments are performed on a simulated robot item retrieval problem and trained in a fully end-to-end manner via imitation learning.	Training end-to-end visual and control networks @cite_17 , has proven difficult due to long roll outs and large action spaces. Within reinforcement learning, several approaches for mapping natural language instructions to actions rely on reward shapping @cite_14 @cite_0 and imitation learning @cite_21 @cite_3 . Imitation learning has also proven effective for fine grained activities like grasping @cite_9 , leading to state-of-the-art results on a broad set of tasks @cite_37 . The difficulty encountered in these scenarios emphasizes the need to explore new methods for efficient learning of multimodal representations. @cite_26 explored attention model architectures, but do not include early fusion techniques. Early fusion of goal information has shown promise with small observation spaces @cite_44 , but our work begins to explore this method for high-dimensional visual domains. In this paper, we hope to provide some insight into this approach and highlight its power in interactive settings.	{ "cite_N": [ "@cite_37", "@cite_14", "@cite_26", "@cite_9", "@cite_21", "@cite_3", "@cite_0", "@cite_44", "@cite_17" ], "mid": [ "2741439293", "2611884151", "2791148095", "2106338164", "2963451844", "2899235916", "2890902815", "2963428623", "2964161785" ], "abstract": [ "", "We propose to directly map raw visual observations and text input to actions for instruction execution. While existing approaches assume access to structured environment representations or use a pipeline of separately trained models, we learn a single model to jointly reason about linguistic and visual input. We use reinforcement learning in a contextual bandit setting to train a neural network agent. To guide the agent's exploration, we use reward shaping with different forms of supervision. Our approach does not require intermediate representations, planning procedures, or training different models. We evaluate in a simulated environment, and show significant improvements over supervised learning and common reinforcement learning variants.", "Visual Question Answering (VQA) is an increasingly popular topic in deep learning research, requiring coordination of natural language processing and computer vision modules into a single architecture. We build upon the model which placed first in the VQA Challenge by developing thirteen new attention mechanisms and introducing a simplified classifier. We performed 300 GPU hours of extensive hyperparameter and architecture searches and were able to achieve an evaluation score of 64.78 , outperforming the existing state-of-the-art single model's validation score of 63.15 .", "In this paper, we present an approach that allows a robot to observe, generalize, and reproduce tasks observed from multiple demonstrations. Motion capture data is recorded in which a human instructor manipulates a set of objects. In our approach, we learn relations between body parts of the demonstrator and objects in the scene. These relations result in a generalized task description. The problem of learning and reproducing human actions is formulated using a dynamic Bayesian network (DBN). The posteriors corresponding to the nodes of the DBN are estimated by observing objects in the scene and body parts of the demonstrator. To reproduce a task, we seek for the maximum-likelihood action sequence according to the DBN. We additionally show how further constraints can be incorporated online, for example, to robustly deal with unforeseen obstacles. Experiments carried out with a real 6-DoF robotic manipulator as well as in simulation show that our approach enables a robot to reproduce a task carried out by a human demonstrator. Our approach yields a high degree of generalization illustrated by performing a pick-and-place and a whiteboard cleaning task.", "", "", "", "We present a learning-based mapless motion planner by taking the sparse 10-dimensional range findings and the target position with respect to the mobile robot coordinate frame as input and the continuous steering commands as output. Traditional motion planners for mobile ground robots with a laser range sensor mostly depend on the obstacle map of the navigation environment where both the highly precise laser sensor and the obstacle map building work of the environment are indispensable. We show that, through an asynchronous deep reinforcement learning method, a mapless motion planner can be trained end-to-end without any manually designed features and prior demonstrations. The trained planner can be directly applied in unseen virtual and real environments. The experiments show that the proposed mapless motion planner can navigate the nonholonomic mobile robot to the desired targets without colliding with any obstacles.", "Policy search methods can allow robots to learn control policies for a wide range of tasks, but practical applications of policy search often require hand-engineered components for perception, state estimation, and low-level control. In this paper, we aim to answer the following question: does training the perception and control systems jointly end-to-end provide better performance than training each component separately? To this end, we develop a method that can be used to learn policies that map raw image observations directly to torques at the robot's motors. The policies are represented by deep convolutional neural networks (CNNs) with 92,000 parameters, and are trained using a guided policy search method, which transforms policy search into supervised learning, with supervision provided by a simple trajectory-centric reinforcement learning method. We evaluate our method on a range of real-world manipulation tasks that require close coordination between vision and control, such as screwing a cap onto a bottle, and present simulated comparisons to a range of prior policy search methods." ] }
1811.08783	2900815132	Many audio signal processing methods are formulated in the time-frequency (T-F) domain which is obtained by the short-time Fourier transform (STFT). The properties of the STFT are fully characterized by window function, number of frequency channels, and time-shift. Thus, designing a better window is important for improving the performance of the processing especially when a less redundant T-F representation is desirable. While many window functions have been proposed in the literature, they are designed to have a good frequency response for analysis, which may not perform well in terms of signal processing. The window design must take the effect of the reconstruction (from the T-F domain into the time domain) into account for improving the performance. In this paper, an optimization-based design method of a nearly tight window is proposed to obtain a window performing well for the T-F domain signal processing.	For designing a low-condition-numbered window, design methods of tight windows have been proposed @cite_28 @cite_20 . These methods aim to find a tight window with better frequency responses. However, since the constraint to the tight window greatly limits the set of variables, desired characteristics may not be obtained.	{ "cite_N": [ "@cite_28", "@cite_20" ], "mid": [ "1993505515", "2963293266" ], "abstract": [ "Weyl-Heisenberg frames are a principal tool of short-time Fourier analysis. We present a comprehensive study of Weyl-Heisenberg frames in l sup 2 (Z), with a focus on frames that are tight. A number of properties of these frames are derived. A complete parameterization of finite-length windows for tight Weyl-Heisenberg frames in l sup 2 (Z) is described. Design of windows for tight Weyl-Heisenberg frames requires optimization of their frequency characteristics under nonlinear constraints. We propose an efficient design method based on expansions with respect to prolate spheroidal sequences. The advantages of the proposed method over standard optimization procedures include a reduction in computational complexity and the ability to provide long windows that can be specified concisely using only a few parameters; these advantages become increasingly pronounced as the frame redundancy increases. The resilience of overcomplete Weyl-Heisenberg expansions to additive noise and quantization is also studied. We show that manifestations of degradation due to uncorrelated zero-mean additive noise are inversely proportional to the expansion redundancy, whereas the quantization error is for a given quantization step inversely proportional to the square of the expansion redundancy.", "Redundant Gabor frames admit an infinite number of dual frames, yet only the canonical dual Gabor system, constructed from the minimal l2-norm dual window, is widely used. This window function however, might lack desirable properties, e.g. good time–frequency concentration, small support or smoothness. We employ convex optimization methods to design dual windows satisfying the Wexler–Raz equations and optimizing various constraints. Numerical experiments suggest that alternate dual windows with considerably improved features can be found." ] }
1811.08783	2900815132	Many audio signal processing methods are formulated in the time-frequency (T-F) domain which is obtained by the short-time Fourier transform (STFT). The properties of the STFT are fully characterized by window function, number of frequency channels, and time-shift. Thus, designing a better window is important for improving the performance of the processing especially when a less redundant T-F representation is desirable. While many window functions have been proposed in the literature, they are designed to have a good frequency response for analysis, which may not perform well in terms of signal processing. The window design must take the effect of the reconstruction (from the T-F domain into the time domain) into account for improving the performance. In this paper, an optimization-based design method of a nearly tight window is proposed to obtain a window performing well for the T-F domain signal processing.	On the other hand, some methods of nearly-tight window design have been proposed @cite_24 @cite_29 @cite_1 . One approach of this research is to minimize the difference between the frame operator and identity operator using the gradient-based optimization @cite_24 @cite_29 . These methods minimize the distance to the set of tight windows by the gradient method, whereby they have a possibility of falling into the local minima. Another approach is to replace the non-convex cost of measuring the distance to the tight window with convex functions @cite_1 . Since that method is formulated as convex optimization, it is guaranteed that globally optimal solutions can be obtained, so a trade-off between the condition number and the frequency response can be easily considered. However, as a result of approximating the cost function, the obtained solutions may not be close to the original solution which is tight. The cost should be reduced strictly without approximation, while the trade-off should be easily adjusted.	{ "cite_N": [ "@cite_24", "@cite_29", "@cite_1" ], "mid": [ "2105052861", "", "2096029268" ], "abstract": [ "Designing optimal perfect-reconstruction (PR) and near PR (NPR) cosine-modulated filter banks is essentially a constrained nonlinear minimization problem. This paper proposes two second-order cone-programming based algorithms for designing NPR and practically PR cosine-modulated filter banks with improved performance relative to several established design methods.", "", "We propose a flexible, efficient design technique for the prototype filter of an oversampled near perfect reconstruction (NPR) generalized discrete Fourier transform (GDFT) filterbank. Such filterbanks have several desirable properties for subband processing systems that are sensitive to aliasing, such as subband adaptive filters. The design criteria for the prototype filter are explicit bounds (derived herein) on the aliased components in the subbands and the output, the distortion induced by the filterbank, and the imaged subband errors in the output. It is shown that the design of an optimal prototype filter can be transformed into a convex optimization problem, which can be efficiently solved. The proposed design technique provides an efficient and effective tool for exploring many of the inherent tradeoffs in the design of the prototype filter, including the tradeoff between aliasing in the subbands and the distortion induced by the filterbank. We calculate several examples of these tradeoffs and demonstrate that the proposed method can generate filters with significantly better performance than filters obtained using current design methods." ] }
1811.08747	2963152299	Image dehazing aims to recover the uncorrupted content from a hazy image. Instead of leveraging traditional low-level or handcrafted image priors as the restoration constraints, e.g., dark channels and increased contrast, we propose an end-to-end gated context aggregation network to directly restore the final haze-free image. In this network, we adopt the latest smoothed dilation technique to help remove the gridding artifacts caused by the widely-used dilated convolution with negligible extra parameters, and leverage a gated sub-network to fuse the features from different levels. Extensive experiments demonstrate that our method can surpass previous state-of-the-art methods by a large margin both quantitatively and qualitatively. In addition, to demonstrate the generality of the proposed method, we further apply it to the image deraining task, which also achieves the state-of-the-art performance.	In the traditional prior-based methods, many different image statistics priors are leveraged as extra constraints to compensate for the information loss during the corruption procedure. For example, @cite_22 propose a physically grounded method by estimating the albedo of the scene. @cite_18 @cite_43 @cite_34 discover and improve the effective dark channel prior to calculate the intermediate transmission map more reliably. @cite_31 use Markov Random Field to maximize the local contrast of an image by assuming that the local contrast of a clear image is higher than that of a hazy image. Based on the observation that small image patches typically exhibit a one-dimensional distribution in the RGB color space, @cite_0 recently propose a color-line method for image dehazing and @cite_1 propose a non-local path prior to characterize the clean images. These dedicatedly handcrafted priors , however, hold for some cases, but they are not always robust to handle all the cases.	{ "cite_N": [ "@cite_18", "@cite_22", "@cite_1", "@cite_0", "@cite_43", "@cite_31", "@cite_34" ], "mid": [ "2128254161", "2028990532", "", "2028763589", "2106402996", "2114867966", "2058843488" ], "abstract": [ "In this paper, we propose a simple but effective image prior-dark channel prior to remove haze from a single input image. The dark channel prior is a kind of statistics of outdoor haze-free images. It is based on a key observation-most local patches in outdoor haze-free images contain some pixels whose intensity is very low in at least one color channel. Using this prior with the haze imaging model, we can directly estimate the thickness of the haze and recover a high-quality haze-free image. Results on a variety of hazy images demonstrate the power of the proposed prior. Moreover, a high-quality depth map can also be obtained as a byproduct of haze removal.", "In this paper we present a new method for estimating the optical transmission in hazy scenes given a single input image. Based on this estimation, the scattered light is eliminated to increase scene visibility and recover haze-free scene contrasts. In this new approach we formulate a refined image formation model that accounts for surface shading in addition to the transmission function. This allows us to resolve ambiguities in the data by searching for a solution in which the resulting shading and transmission functions are locally statistically uncorrelated. A similar principle is used to estimate the color of the haze. Results demonstrate the new method abilities to remove the haze layer as well as provide a reliable transmission estimate which can be used for additional applications such as image refocusing and novel view synthesis.", "", "Photographs of hazy scenes typically suffer having low contrast and offer a limited visibility of the scene. This article describes a new method for single-image dehazing that relies on a generic regularity in natural images where pixels of small image patches typically exhibit a 1D distribution in RGB color space, known as color-lines. We derive a local formation model that explains the color-lines in the context of hazy scenes and use it for recovering the scene transmission based on the lines' offset from the origin. The lack of a dominant color-line inside a patch or its lack of consistency with the formation model allows us to identify and avoid false predictions. Thus, unlike existing approaches that follow their assumptions across the entire image, our algorithm validates its hypotheses and obtains more reliable estimates where possible. In addition, we describe a Markov random field model dedicated to producing complete and regularized transmission maps given noisy and scattered estimates. Unlike traditional field models that consist of local coupling, the new model is augmented with long-range connections between pixels of similar attributes. These connections allow our algorithm to properly resolve the transmission in isolated regions where nearby pixels do not offer relevant information. An extensive evaluation of our method over different types of images and its comparison to state-of-the-art methods over established benchmark images show a consistent improvement in the accuracy of the estimated scene transmission and recovered haze-free radiances.", "In this paper, we propose an improved image dehazing algorithm using dark channel prior and Multi-Scale Retinex. Main improvement lies in automatic and fast acquisition of transmission map of the scene. We implement the Multi-scale Retinex algorithm on the luminance component in YCbCr space, obtain a pseudo transmission map whose function is similar to the transmission map in original approach. Combining with the haze image model and the dark channel prior, we can recover a high quality haze-free image. Compared with the original method, our algorithm has two main advantages: (i) no user interaction is needed, and (ii) restoring the image much faster while maintaining comparable dehazing performance.", "Bad weather, such as fog and haze, can significantly degrade the visibility of a scene. Optically, this is due to the substantial presence of particles in the atmosphere that absorb and scatter light. In computer vision, the absorption and scattering processes are commonly modeled by a linear combination of the direct attenuation and the airlight. Based on this model, a few methods have been proposed, and most of them require multiple input images of a scene, which have either different degrees of polarization or different atmospheric conditions. This requirement is the main drawback of these methods, since in many situations, it is difficult to be fulfilled. To resolve the problem, we introduce an automated method that only requires a single input image. This method is based on two basic observations: first, images with enhanced visibility (or clear-day images) have more contrast than images plagued by bad weather; second, airlight whose variation mainly depends on the distance of objects to the viewer, tends to be smooth. Relying on these two observations, we develop a cost function in the framework of Markov random fields, which can be efficiently optimized by various techniques, such as graph-cuts or belief propagation. The method does not require the geometrical information of the input image, and is applicable for both color and gray images.", "In the frog and haze climatic condition, the captured picture will become blurred and the color is partial gray and white, due to the effect of atmospheric scattering. This situation brings a great deal of inconvenience to the video surveillance system, so the study of defogging algorithm in this weather is of great importance. This paper deeply analyzes the physical process of imaging in foggy weather. After full study on the haze removal algorithm of single image over the last decade, we propo se a fast haze removal algorithm which based on a fast bilateral filtering combined with dark colors prior. This algorithm starts with the atmospheric scattering model, derives a estimated transmission map by using dark channel prior, and then combines with grayscale to extract refined transmission map by using the fast bilateral filter. This algorithm has a fast execution speed and greatly improves the original algorithm which is morre time-consuming. On this basis, we analyzed the reasons why the image is dim after the haze removal using dark channel prior, and proposed the improved transmission map formula. Experimental-results show that this algorithm is feasible which effectively restores the contrast and color of the scene, significantly improves the visual effects of the image. Those image with large area of sky usually prone to distortion when using the dark channel prior, Therefore we propose a method of weakening the sky region, aims to improve the adaptability of the algorithm." ] }
1811.08747	2963152299	Image dehazing aims to recover the uncorrupted content from a hazy image. Instead of leveraging traditional low-level or handcrafted image priors as the restoration constraints, e.g., dark channels and increased contrast, we propose an end-to-end gated context aggregation network to directly restore the final haze-free image. In this network, we adopt the latest smoothed dilation technique to help remove the gridding artifacts caused by the widely-used dilated convolution with negligible extra parameters, and leverage a gated sub-network to fuse the features from different levels. Extensive experiments demonstrate that our method can surpass previous state-of-the-art methods by a large margin both quantitatively and qualitatively. In addition, to demonstrate the generality of the proposed method, we further apply it to the image deraining task, which also achieves the state-of-the-art performance.	Recently, learning-based methods are proposed for image dehazing by leveraging the large-scale datasets and the powerful parallelism of GPU. In these type of methods, the image priors are automatically learned from the training dataset by the neural network and saved in the network weights. Their main differences typically lie in the learning targets and the detailed network structures. @cite_20 @cite_45 propose an end-to-end CNN network and multi-scale network respectively to predict the intermediate transmission maps. However, inaccuracies in the estimation of the transmission map always lead to low-quality dehazed results. @cite_40 encode the transmission map and the atmospheric light into one variable, and then use a lightweight network to predict it. @cite_7 design two different sub-networks for the prediction of the transmission map and the atmospheric light by following the physical model defined in eq:dehaze_model . We propose an end-to-end gated context aggregation network for image dehazing but different from these methods, our proposed is designed to directly regress the residue between the hazy image and the target clean image. Moreover, our network structure definitely distinguish from the previous ones, which is quite lightweight but can achieve much better results than all the previous methods.	{ "cite_N": [ "@cite_40", "@cite_45", "@cite_7", "@cite_20" ], "mid": [ "", "2519481857", "2963306157", "2256362396" ], "abstract": [ "", "The performance of existing image dehazing methods is limited by hand-designed features, such as the dark channel, color disparity and maximum contrast, with complex fusion schemes. In this paper, we propose a multi-scale deep neural network for single-image dehazing by learning the mapping between hazy images and their corresponding transmission maps. The proposed algorithm consists of a coarse-scale net which predicts a holistic transmission map based on the entire image, and a fine-scale net which refines results locally. To train the multi-scale deep network, we synthesize a dataset comprised of hazy images and corresponding transmission maps based on the NYU Depth dataset. Extensive experiments demonstrate that the proposed algorithm performs favorably against the state-of-the-art methods on both synthetic and real-world images in terms of quality and speed.", "We propose a new end-to-end single image dehazing method, called Densely Connected Pyramid Dehazing Network (DCPDN), which can jointly learn the transmission map, atmospheric light and dehazing all together. The end-to-end learning is achieved by directly embedding the atmospheric scattering model into the network, thereby ensuring that the proposed method strictly follows the physics-driven scattering model for dehazing. Inspired by the dense network that can maximize the information flow along features from different levels, we propose a new edge-preserving densely connected encoder-decoder structure with multi-level pyramid pooling module for estimating the transmission map. This network is optimized using a newly introduced edge-preserving loss function. To further incorporate the mutual structural information between the estimated transmission map and the dehazed result, we propose a joint-discriminator based on generative adversarial network framework to decide whether the corresponding dehazed image and the estimated transmission map are real or fake. An ablation study is conducted to demonstrate the effectiveness of each module evaluated at both estimated transmission map and dehazed result. Extensive experiments demonstrate that the proposed method achieves significant improvements over the state-of-the-art methods. Code and dataset is made available at: https: github.com hezhangsprinter DCPDN", "Single image haze removal is a challenging ill-posed problem. Existing methods use various constraints priors to get plausible dehazing solutions. The key to achieve haze removal is to estimate a medium transmission map for an input hazy image. In this paper, we propose a trainable end-to-end system called DehazeNet, for medium transmission estimation. DehazeNet takes a hazy image as input, and outputs its medium transmission map that is subsequently used to recover a haze-free image via atmospheric scattering model. DehazeNet adopts convolutional neural network-based deep architecture, whose layers are specially designed to embody the established assumptions priors in image dehazing. Specifically, the layers of Maxout units are used for feature extraction, which can generate almost all haze-relevant features. We also propose a novel nonlinear activation function in DehazeNet, called bilateral rectified linear unit, which is able to improve the quality of recovered haze-free image. We establish connections between the components of the proposed DehazeNet and those used in existing methods. Experiments on benchmark images show that DehazeNet achieves superior performance over existing methods, yet keeps efficient and easy to use." ] }
1811.08610	2900616773	Machine Reading Comprehension (MRC) with multiple-choice questions requires the machine to read given passage and select the correct answer among several candidates. In this paper, we propose a novel approach called Convolutional Spatial Attention (CSA) model which can better handle the MRC with multiple-choice questions. The proposed model could fully extract the mutual information among the passage, question, and the candidates, to form the enriched representations. Furthermore, to merge various attention results, we propose to use convolutional operation to dynamically summarize the attention values within the different size of regions. Experimental results show that the proposed model could give substantial improvements over various state-of-the-art systems on both RACE and SemEval-2018 Task11 datasets.	Massive progress has been made on machine reading comprehension field in recent years. The booming of the MRC can trace back to the release of the large-scale datasets, such as CNN DailyMail @cite_8 and CBT @cite_26 ). After the release of these datasets, various neural network approaches @cite_17 @cite_7 @cite_0 @cite_24 have been proposed and become fundamental components in the future studies. Another representative dataset is SQuAD @cite_19 , which was difficult than the cloze-style reading comprehension and requires the machine to generate a span in the passage to answer the questions. With rapid progress on designing effective neural network models @cite_25 @cite_12 @cite_21 @cite_22 , recent works on this datasets have surpassed the average human performance, such as QANet @cite_15 etc.	{ "cite_N": [ "@cite_26", "@cite_22", "@cite_7", "@cite_8", "@cite_21", "@cite_0", "@cite_24", "@cite_19", "@cite_15", "@cite_25", "@cite_12", "@cite_17" ], "mid": [ "2126209950", "2752104716", "2288995089", "2949615363", "2740747242", "2516196286", "2415755012", "2427527485", "2798858969", "2552027021", "2951815760", "2411480514" ], "abstract": [ "We introduce a new test of how well language models capture meaning in children's books. Unlike standard language modelling benchmarks, it distinguishes the task of predicting syntactic function words from that of predicting lower-frequency words, which carry greater semantic content. We compare a range of state-of-the-art models, each with a different way of encoding what has been previously read. We show that models which store explicit representations of long-term contexts outperform state-of-the-art neural language models at predicting semantic content words, although this advantage is not observed for syntactic function words. Interestingly, we find that the amount of text encoded in a single memory representation is highly influential to the performance: there is a sweet-spot, not too big and not too small, between single words and full sentences that allows the most meaningful information in a text to be effectively retained and recalled. Further, the attention over such window-based memories can be trained effectively through self-supervision. We then assess the generality of this principle by applying it to the CNN QA benchmark, which involves identifying named entities in paraphrased summaries of news articles, and achieve state-of-the-art performance.", "In this paper, we introduce the Reinforced Mnemonic Reader for machine comprehension (MC) task, which aims to answer a query about a given context document. We propose several novel mechanisms that address critical problems in MC that are not adequately solved by previous works, such as enhancing the capacity of encoder, modeling long-term dependencies of contexts, refining the predicted answer span, and directly optimizing the evaluation metric. Extensive experiments on TriviaQA and Stanford Question Answering Dataset (SQuAD) show that our model achieves state-of-the-art results.", "Several large cloze-style context-question-answer datasets have been introduced recently: the CNN and Daily Mail news data and the Children's Book Test. Thanks to the size of these datasets, the associated text comprehension task is well suited for deep-learning techniques that currently seem to outperform all alternative approaches. We present a new, simple model that uses attention to directly pick the answer from the context as opposed to computing the answer using a blended representation of words in the document as is usual in similar models. This makes the model particularly suitable for question-answering problems where the answer is a single word from the document. Ensemble of our models sets new state of the art on all evaluated datasets.", "Teaching machines to read natural language documents remains an elusive challenge. Machine reading systems can be tested on their ability to answer questions posed on the contents of documents that they have seen, but until now large scale training and test datasets have been missing for this type of evaluation. In this work we define a new methodology that resolves this bottleneck and provides large scale supervised reading comprehension data. This allows us to develop a class of attention based deep neural networks that learn to read real documents and answer complex questions with minimal prior knowledge of language structure.", "", "Cloze-style queries are representative problems in reading comprehension. Over the past few months, we have seen much progress that utilizing neural network approach to solve Cloze-style questions. In this paper, we present a novel model called attention-over-attention reader for the Cloze-style reading comprehension task. Our model aims to place another attention mechanism over the document-level attention, and induces \"attended attention\" for final predictions. Unlike the previous works, our neural network model requires less pre-defined hyper-parameters and uses an elegant architecture for modeling. Experimental results show that the proposed attention-over-attention model significantly outperforms various state-of-the-art systems by a large margin in public datasets, such as CNN and Children's Book Test datasets.", "In this paper we study the problem of answering cloze-style questions over documents. Our model, the Gated-Attention (GA) Reader, integrates a multi-hop architecture with a novel attention mechanism, which is based on multiplicative interactions between the query embedding and the intermediate states of a recurrent neural network document reader. This enables the reader to build query-specific representations of tokens in the document for accurate answer selection. The GA Reader obtains state-of-the-art results on three benchmarks for this task--the CNN & Daily Mail news stories and the Who Did What dataset. The effectiveness of multiplicative interaction is demonstrated by an ablation study, and by comparing to alternative compositional operators for implementing the gated-attention. The code is available at this https URL", "We present the Stanford Question Answering Dataset (SQuAD), a new reading comprehension dataset consisting of 100,000+ questions posed by crowdworkers on a set of Wikipedia articles, where the answer to each question is a segment of text from the corresponding reading passage. We analyze the dataset to understand the types of reasoning required to answer the questions, leaning heavily on dependency and constituency trees. We build a strong logistic regression model, which achieves an F1 score of 51.0 , a significant improvement over a simple baseline (20 ). However, human performance (86.8 ) is much higher, indicating that the dataset presents a good challenge problem for future research. The dataset is freely available at this https URL", "Current end-to-end machine reading and question answering (Q &A) models are primarily based on recurrent neural networks (RNNs) with attention. Despite their success, these models are often slow for both training and inference due to the sequential nature of RNNs. We propose a new Q &A architecture called QANet, which does not require recurrent networks: Its encoder consists exclusively of convolution and self-attention, where convolution models local interactions and self-attention models global interactions. On the SQuAD dataset, our model is 3x to 13x faster in training and 4x to 9x faster in inference, while achieving equivalent accuracy to recurrent models. The speed-up gain allows us to train the model with much more data. We hence combine our model with data generated by backtranslation from a neural machine translation model. On the SQuAD dataset, our single model, trained with augmented data, achieves 84.6 F1 score on the test set, which is significantly better than the best published F1 score of 81.8.", "Several deep learning models have been proposed for question answering. However, due to their single-pass nature, they have no way to recover from local maxima corresponding to incorrect answers. To address this problem, we introduce the Dynamic Coattention Network (DCN) for question answering. The DCN first fuses co-dependent representations of the question and the document in order to focus on relevant parts of both. Then a dynamic pointing decoder iterates over potential answer spans. This iterative procedure enables the model to recover from initial local maxima corresponding to incorrect answers. On the Stanford question answering dataset, a single DCN model improves the previous state of the art from 71.0 F1 to 75.9 , while a DCN ensemble obtains 80.4 F1.", "Machine comprehension (MC), answering a query about a given context paragraph, requires modeling complex interactions between the context and the query. Recently, attention mechanisms have been successfully extended to MC. Typically these methods use attention to focus on a small portion of the context and summarize it with a fixed-size vector, couple attentions temporally, and or often form a uni-directional attention. In this paper we introduce the Bi-Directional Attention Flow (BIDAF) network, a multi-stage hierarchical process that represents the context at different levels of granularity and uses bi-directional attention flow mechanism to obtain a query-aware context representation without early summarization. Our experimental evaluations show that our model achieves the state-of-the-art results in Stanford Question Answering Dataset (SQuAD) and CNN DailyMail cloze test.", "Enabling a computer to understand a document so that it can answer comprehension questions is a central, yet unsolved goal of NLP. A key factor impeding its solution by machine learned systems is the limited availability of human-annotated data. (2015) seek to solve this problem by creating over a million training examples by pairing CNN and Daily Mail news articles with their summarized bullet points, and show that a neural network can then be trained to give good performance on this task. In this paper, we conduct a thorough examination of this new reading comprehension task. Our primary aim is to understand what depth of language understanding is required to do well on this task. We approach this from one side by doing a careful hand-analysis of a small subset of the problems and from the other by showing that simple, carefully designed systems can obtain accuracies of 73.6 and 76.6 on these two datasets, exceeding current state-of-the-art results by 7-10 and approaching what we believe is the ceiling for performance on this task." ] }
1811.08610	2900616773	Machine Reading Comprehension (MRC) with multiple-choice questions requires the machine to read given passage and select the correct answer among several candidates. In this paper, we propose a novel approach called Convolutional Spatial Attention (CSA) model which can better handle the MRC with multiple-choice questions. The proposed model could fully extract the mutual information among the passage, question, and the candidates, to form the enriched representations. Furthermore, to merge various attention results, we propose to use convolutional operation to dynamically summarize the attention values within the different size of regions. Experimental results show that the proposed model could give substantial improvements over various state-of-the-art systems on both RACE and SemEval-2018 Task11 datasets.	However, current machine reading comprehension models are still struggling with solving the questions that need reasoning over multiple sentences or even passage. To solve the reading comprehension with multiple-choice questions, various approaches have been proposed, and most of them are focusing on designing effective attentions or persuing enriched representations for prediction. When releasing the RACE dataset, lai2017race also adopted and modified two models of the cloze-style reading comprehension: Gated Attention Reader @cite_24 and Stanford Attentive Reader @cite_17 . However, experimental results show that these models are not capable of this task. parikh2018eliminet introduced ElimiNet which use a combination of elimination and selection to get refined representation of the candidates. Xu2017Towards proposed the Dynamic Fusion Networks (DFN), which uses multi-hop reasoning mechanism for this task. AAAI1816331 proposed the Hierarchical Attention Flow model, which leverage candidate options to model the interactions among passage, questions, and candidates.	{ "cite_N": [ "@cite_24", "@cite_17" ], "mid": [ "2415755012", "2411480514" ], "abstract": [ "In this paper we study the problem of answering cloze-style questions over documents. Our model, the Gated-Attention (GA) Reader, integrates a multi-hop architecture with a novel attention mechanism, which is based on multiplicative interactions between the query embedding and the intermediate states of a recurrent neural network document reader. This enables the reader to build query-specific representations of tokens in the document for accurate answer selection. The GA Reader obtains state-of-the-art results on three benchmarks for this task--the CNN & Daily Mail news stories and the Who Did What dataset. The effectiveness of multiplicative interaction is demonstrated by an ablation study, and by comparing to alternative compositional operators for implementing the gated-attention. The code is available at this https URL", "Enabling a computer to understand a document so that it can answer comprehension questions is a central, yet unsolved goal of NLP. A key factor impeding its solution by machine learned systems is the limited availability of human-annotated data. (2015) seek to solve this problem by creating over a million training examples by pairing CNN and Daily Mail news articles with their summarized bullet points, and show that a neural network can then be trained to give good performance on this task. In this paper, we conduct a thorough examination of this new reading comprehension task. Our primary aim is to understand what depth of language understanding is required to do well on this task. We approach this from one side by doing a careful hand-analysis of a small subset of the problems and from the other by showing that simple, carefully designed systems can obtain accuracies of 73.6 and 76.6 on these two datasets, exceeding current state-of-the-art results by 7-10 and approaching what we believe is the ceiling for performance on this task." ] }
1906.09072	2952282509	Many studies have been done to prove the vulnerability of neural networks to adversarial example. A trained and well-behaved model can be fooled by a visually imperceptible perturbation, i.e., an originally correctly classified image could be misclassified after a slight perturbation. In this paper, we propose a black-box strategy to attack such networks using an evolution algorithm. First, we formalize the generation of an adversarial example into the optimization problem of perturbations that represent the noise added to an original image at each pixel. To solve this optimization problem in a black-box way, we find that an evolution algorithm perfectly meets our requirement since it can work without any gradient information. Therefore, we test various evolution algorithms, including a simple genetic algorithm, a parameter-exploring policy gradient, an OpenAI evolution strategy, and a covariance matrix adaptive evolution strategy. Experimental results show that a covariance matrix adaptive evolution Strategy performs best in this optimization problem. Additionally, we also perform several experiments to explore the effect of different regularizations on improving the quality of an adversarial example.	Although DCNNs have been proven to be a powerful learning model, @cite_3 first found that a hardly perceptible perturbation would lead to misclassification of the trained Network. @cite_10 applied the FGS algorithm to generate an adversarial example and explained that the linear nature of DCNNs was responsible for its vulnerability:	{ "cite_N": [ "@cite_10", "@cite_3" ], "mid": [ "2963207607", "2964153729" ], "abstract": [ "Abstract: Several machine learning models, including neural networks, consistently misclassify adversarial examples---inputs formed by applying small but intentionally worst-case perturbations to examples from the dataset, such that the perturbed input results in the model outputting an incorrect answer with high confidence. Early attempts at explaining this phenomenon focused on nonlinearity and overfitting. We argue instead that the primary cause of neural networks' vulnerability to adversarial perturbation is their linear nature. This explanation is supported by new quantitative results while giving the first explanation of the most intriguing fact about them: their generalization across architectures and training sets. Moreover, this view yields a simple and fast method of generating adversarial examples. Using this approach to provide examples for adversarial training, we reduce the test set error of a maxout network on the MNIST dataset.", "Abstract: Deep neural networks are highly expressive models that have recently achieved state of the art performance on speech and visual recognition tasks. While their expressiveness is the reason they succeed, it also causes them to learn uninterpretable solutions that could have counter-intuitive properties. In this paper we report two such properties. First, we find that there is no distinction between individual high level units and random linear combinations of high level units, according to various methods of unit analysis. It suggests that it is the space, rather than the individual units, that contains of the semantic information in the high layers of neural networks. Second, we find that deep neural networks learn input-output mappings that are fairly discontinuous to a significant extend. We can cause the network to misclassify an image by applying a certain imperceptible perturbation, which is found by maximizing the network's prediction error. In addition, the specific nature of these perturbations is not a random artifact of learning: the same perturbation can cause a different network, that was trained on a different subset of the dataset, to misclassify the same input." ] }
1906.09072	2952282509	Many studies have been done to prove the vulnerability of neural networks to adversarial example. A trained and well-behaved model can be fooled by a visually imperceptible perturbation, i.e., an originally correctly classified image could be misclassified after a slight perturbation. In this paper, we propose a black-box strategy to attack such networks using an evolution algorithm. First, we formalize the generation of an adversarial example into the optimization problem of perturbations that represent the noise added to an original image at each pixel. To solve this optimization problem in a black-box way, we find that an evolution algorithm perfectly meets our requirement since it can work without any gradient information. Therefore, we test various evolution algorithms, including a simple genetic algorithm, a parameter-exploring policy gradient, an OpenAI evolution strategy, and a covariance matrix adaptive evolution strategy. Experimental results show that a covariance matrix adaptive evolution Strategy performs best in this optimization problem. Additionally, we also perform several experiments to explore the effect of different regularizations on improving the quality of an adversarial example.	@cite_19 improved the FGS and proposed the FGV algorithm: Moosavi- @cite_2 suggested an iteration method named DeepFool to minimize the perturbation @math , which is sufficient to change the predicted result. The existence of universal perturbations was shown in @cite_14 . @cite_15 proposed the adversarial transformation network, which is similar to an autoencoder, to generate an adversarial example. @cite_6 developed an adversarial example for semantic segmentation and object detection.	{ "cite_N": [ "@cite_14", "@cite_6", "@cite_19", "@cite_2", "@cite_15" ], "mid": [ "2543927648", "2604505099", "2963098487", "2243397390", "2604147826" ], "abstract": [ "Given a state-of-the-art deep neural network classifier, we show the existence of a universal (image-agnostic) and very small perturbation vector that causes natural images to be misclassified with high probability. We propose a systematic algorithm for computing universal perturbations, and show that state-of-the-art deep neural networks are highly vulnerable to such perturbations, albeit being quasi-imperceptible to the human eye. We further empirically analyze these universal perturbations and show, in particular, that they generalize very well across neural networks. The surprising existence of universal perturbations reveals important geometric correlations among the high-dimensional decision boundary of classifiers. It further outlines potential security breaches with the existence of single directions in the input space that adversaries can possibly exploit to break a classifier on most natural images.", "It has been well demonstrated that adversarial examples, i.e., natural images with visually imperceptible perturbations added, cause deep networks to fail on image classification. In this paper, we extend adversarial examples to semantic segmentation and object detection which are much more difficult. Our observation is that both segmentation and detection are based on classifying multiple targets on an image (e.g., the target is a pixel or a receptive field in segmentation, and an object proposal in detection). This inspires us to optimize a loss function over a set of targets for generating adversarial perturbations. Based on this, we propose a novel algorithm named Dense Adversary Generation (DAG), which applies to the state-of-the-art networks for segmentation and detection. We find that the adversarial perturbations can be transferred across networks with different training data, based on different architectures, and even for different recognition tasks. In particular, the transfer ability across networks with the same architecture is more significant than in other cases. Besides, we show that summing up heterogeneous perturbations often leads to better transfer performance, which provides an effective method of black-box adversarial attack.", "State-of-the-art deep neural networks suffer from a fundamental problem – they misclassify adversarial examples formed by applying small perturbations to inputs. In this paper, we present a new psychometric perceptual adversarial similarity score (PASS) measure for quantifying adversarial images, introduce the notion of hard positive generation, and use a diverse set of adversarial perturbations – not just the closest ones – for data augmentation. We introduce a novel hot cold approach for adversarial example generation, which provides multiple possible adversarial perturbations for every single image. The perturbations generated by our novel approach often correspond to semantically meaningful image structures, and allow greater flexibility to scale perturbation-amplitudes, which yields an increased diversity of adversarial images. We present adversarial images on several network topologies and datasets, including LeNet on the MNIST dataset, and GoogLeNet and ResidualNet on the ImageNet dataset. Finally, we demonstrate on LeNet and GoogLeNet that fine-tuning with a diverse set of hard positives improves the robustness of these networks compared to training with prior methods of generating adversarial images.", "State-of-the-art deep neural networks have achieved impressive results on many image classification tasks. However, these same architectures have been shown to be unstable to small, well sought, perturbations of the images. Despite the importance of this phenomenon, no effective methods have been proposed to accurately compute the robustness of state-of-the-art deep classifiers to such perturbations on large-scale datasets. In this paper, we fill this gap and propose the DeepFool algorithm to efficiently compute perturbations that fool deep networks, and thus reliably quantify the robustness of these classifiers. Extensive experimental results show that our approach outperforms recent methods in the task of computing adversarial perturbations and making classifiers more robust.1", "Multiple different approaches of generating adversarial examples have been proposed to attack deep neural networks. These approaches involve either directly computing gradients with respect to the image pixels, or directly solving an optimization on the image pixels. In this work, we present a fundamentally new method for generating adversarial examples that is fast to execute and provides exceptional diversity of output. We efficiently train feed-forward neural networks in a self-supervised manner to generate adversarial examples against a target network or set of networks. We call such a network an Adversarial Transformation Network (ATN). ATNs are trained to generate adversarial examples that minimally modify the classifier's outputs given the original input, while constraining the new classification to match an adversarial target class. We present methods to train ATNs and analyze their effectiveness targeting a variety of MNIST classifiers as well as the latest state-of-the-art ImageNet classifier Inception ResNet v2." ] }
1906.09072	2952282509	Many studies have been done to prove the vulnerability of neural networks to adversarial example. A trained and well-behaved model can be fooled by a visually imperceptible perturbation, i.e., an originally correctly classified image could be misclassified after a slight perturbation. In this paper, we propose a black-box strategy to attack such networks using an evolution algorithm. First, we formalize the generation of an adversarial example into the optimization problem of perturbations that represent the noise added to an original image at each pixel. To solve this optimization problem in a black-box way, we find that an evolution algorithm perfectly meets our requirement since it can work without any gradient information. Therefore, we test various evolution algorithms, including a simple genetic algorithm, a parameter-exploring policy gradient, an OpenAI evolution strategy, and a covariance matrix adaptive evolution strategy. Experimental results show that a covariance matrix adaptive evolution Strategy performs best in this optimization problem. Additionally, we also perform several experiments to explore the effect of different regularizations on improving the quality of an adversarial example.	The above works all concern a white-box attack. Considering a black-box attack is more dangerous in the real world, @cite_7 proposed the black-box attack strategy, in which the substitute model needs to be trained using the outputs of the original model as labels. Thus, an adversarial example based on detailed information of the new substitute model can also fool the original model.	{ "cite_N": [ "@cite_7" ], "mid": [ "2603766943" ], "abstract": [ "Machine learning (ML) models, e.g., deep neural networks (DNNs), are vulnerable to adversarial examples: malicious inputs modified to yield erroneous model outputs, while appearing unmodified to human observers. Potential attacks include having malicious content like malware identified as legitimate or controlling vehicle behavior. Yet, all existing adversarial example attacks require knowledge of either the model internals or its training data. We introduce the first practical demonstration of an attacker controlling a remotely hosted DNN with no such knowledge. Indeed, the only capability of our black-box adversary is to observe labels given by the DNN to chosen inputs. Our attack strategy consists in training a local model to substitute for the target DNN, using inputs synthetically generated by an adversary and labeled by the target DNN. We use the local substitute to craft adversarial examples, and find that they are misclassified by the targeted DNN. To perform a real-world and properly-blinded evaluation, we attack a DNN hosted by MetaMind, an online deep learning API. We find that their DNN misclassifies 84.24 of the adversarial examples crafted with our substitute. We demonstrate the general applicability of our strategy to many ML techniques by conducting the same attack against models hosted by Amazon and Google, using logistic regression substitutes. They yield adversarial examples misclassified by Amazon and Google at rates of 96.19 and 88.94 . We also find that this black-box attack strategy is capable of evading defense strategies previously found to make adversarial example crafting harder." ] }
1906.09072	2952282509	Many studies have been done to prove the vulnerability of neural networks to adversarial example. A trained and well-behaved model can be fooled by a visually imperceptible perturbation, i.e., an originally correctly classified image could be misclassified after a slight perturbation. In this paper, we propose a black-box strategy to attack such networks using an evolution algorithm. First, we formalize the generation of an adversarial example into the optimization problem of perturbations that represent the noise added to an original image at each pixel. To solve this optimization problem in a black-box way, we find that an evolution algorithm perfectly meets our requirement since it can work without any gradient information. Therefore, we test various evolution algorithms, including a simple genetic algorithm, a parameter-exploring policy gradient, an OpenAI evolution strategy, and a covariance matrix adaptive evolution strategy. Experimental results show that a covariance matrix adaptive evolution Strategy performs best in this optimization problem. Additionally, we also perform several experiments to explore the effect of different regularizations on improving the quality of an adversarial example.	For adversarial defense, @cite_10 found that adversarial training could reduce network overfitting and, in turn, reinforce the robustness. @cite_9 applied defensive distillation to improve the robustness of the network. @cite_1 tried to use a generative adversarial network (GAN) to train a robust classifier. Its generator is supposed to generate perturbations and its classifier is supposed to correctly classify both the original and perturbed images.	{ "cite_N": [ "@cite_1", "@cite_9", "@cite_10" ], "mid": [ "2612866063", "2964082701", "2963207607" ], "abstract": [ "We propose a novel technique to make neural network robust to adversarial examples using a generative adversarial network. We alternately train both classifier and generator networks. The generator network generates an adversarial perturbation that can easily fool the classifier network by using a gradient of each image. Simultaneously, the classifier network is trained to classify correctly both original and adversarial images generated by the generator. These procedures help the classifier network to become more robust to adversarial perturbations. Furthermore, our adversarial training framework efficiently reduces overfitting and outperforms other regularization methods such as Dropout. We applied our method to supervised learning for CIFAR datasets, and experimantal results show that our method significantly lowers the generalization error of the network. To the best of our knowledge, this is the first method which uses GAN to improve supervised learning.", "Deep learning algorithms have been shown to perform extremely well on manyclassical machine learning problems. However, recent studies have shown thatdeep learning, like other machine learning techniques, is vulnerable to adversarial samples: inputs crafted to force adeep neural network (DNN) to provide adversary-selected outputs. Such attackscan seriously undermine the security of the system supported by the DNN, sometimes with devastating consequences. For example, autonomous vehicles canbe crashed, illicit or illegal content can bypass content filters, or biometricauthentication systems can be manipulated to allow improper access. In thiswork, we introduce a defensive mechanism called defensive distillationto reduce the effectiveness of adversarial samples on DNNs. We analyticallyinvestigate the generalizability and robustness properties granted by the useof defensive distillation when training DNNs. We also empirically study theeffectiveness of our defense mechanisms on two DNNs placed in adversarialsettings. The study shows that defensive distillation can reduce effectivenessof sample creation from 95 to less than 0.5 on a studied DNN. Such dramaticgains can be explained by the fact that distillation leads gradients used inadversarial sample creation to be reduced by a factor of 1030. We alsofind that distillation increases the average minimum number of features thatneed to be modified to create adversarial samples by about 800 on one of theDNNs we tested.", "Abstract: Several machine learning models, including neural networks, consistently misclassify adversarial examples---inputs formed by applying small but intentionally worst-case perturbations to examples from the dataset, such that the perturbed input results in the model outputting an incorrect answer with high confidence. Early attempts at explaining this phenomenon focused on nonlinearity and overfitting. We argue instead that the primary cause of neural networks' vulnerability to adversarial perturbation is their linear nature. This explanation is supported by new quantitative results while giving the first explanation of the most intriguing fact about them: their generalization across architectures and training sets. Moreover, this view yields a simple and fast method of generating adversarial examples. Using this approach to provide examples for adversarial training, we reduce the test set error of a maxout network on the MNIST dataset." ] }
1906.08865	2949936448	This paper presents a technique called evolving self-supervised neural networks - neural networks that can teach themselves, intrinsically motivated, without external supervision or reward. The proposed method presents some sort-of paradigm shift, and differs greatly from both traditional gradient-based learning and evolutionary algorithms in that it combines the metaphor of evolution and learning, more specifically self-learning, together, rather than treating these phenomena alternatively. I simulate a multi-agent system in which neural networks are used to control autonomous foraging agents with little domain knowledge. Experimental results show that only evolved self-supervised agents can demonstrate some sort of intelligent behaviour, but not evolution or self-learning alone. Indications for future work on evolving intelligence are also presented.	Nolfi and his colleagues made a simulation of , or robots, controlled by neural networks situated in a grid-world, with discrete state and action spaces @cite_32 . Each agent lives in its own copy of the world, hence no mutual interaction. The evolutionary task is to evolve action strategies to collect food effectively, while each agent learns to predict the sensory inputs to neural networks for each time step. Learning was implemented using backpropagation based on the error between the actual and the predicted sensory inputs to update the weights of a neural network. It was shown that learning to predict can enhance the evolutionary search, hence increasing the performance of the robot.	{ "cite_N": [ "@cite_32" ], "mid": [ "1979356863" ], "abstract": [ "This article describes simulations on populations of neural networks that both evolve at the population level and learn at the individual level. Unlike other simulations, the evolutionary task (finding food in the environment) and the learning task (predicting the next position of food on the basis of present position and planned network's movement) are different tasks. In these conditions, learning influences evolution (without Lamarckian inheritance of learned weight changes) and evolution influences learning. Average but not peak fitness has a better evolutionary growth with learning than without learning. After the initial generations, individuals that learn to predict during life also improve their food-finding ability during life. Furthermore, individuals that inherit an innate capacity to find food also inherit an innate predisposition to learn to predict the sensory consequences of their movements. They do not predict better at birth, but they do learn to predict better than individuals of the ini..." ] }
1906.08865	2949936448	This paper presents a technique called evolving self-supervised neural networks - neural networks that can teach themselves, intrinsically motivated, without external supervision or reward. The proposed method presents some sort-of paradigm shift, and differs greatly from both traditional gradient-based learning and evolutionary algorithms in that it combines the metaphor of evolution and learning, more specifically self-learning, together, rather than treating these phenomena alternatively. I simulate a multi-agent system in which neural networks are used to control autonomous foraging agents with little domain knowledge. Experimental results show that only evolved self-supervised agents can demonstrate some sort of intelligent behaviour, but not evolution or self-learning alone. Indications for future work on evolving intelligence are also presented.	Scientists at DeepMind recently also used the same idea of the Baldwin Effect, using a genetic algorithm to evolve the initial weights for deep neural networks @cite_6 . By combining the advantage of searching over a vast distribution of weights of evolutionary search and the exploitative power of a gradient descent learning, they reported a that can solve a multiple tasks including regression and physical robot environments. This result is an indication to create meta-learning, another step towards AGI.	{ "cite_N": [ "@cite_6" ], "mid": [ "2808809512" ], "abstract": [ "The scope of the Baldwin effect was recently called into question by two papers that closely examined the seminal work of Hinton and Nowlan. To this date there has been no demonstration of its necessity in empirically challenging tasks. Here we show that the Baldwin effect is capable of evolving few-shot supervised and reinforcement learning mechanisms, by shaping the hyperparameters and the initial parameters of deep learning algorithms. Furthermore it can genetically accommodate strong learning biases on the same set of problems as a recent machine learning algorithm called MAML \"Model Agnostic Meta-Learning\" which uses second-order gradients instead of evolution to learn a set of reference parameters (initial weights) that can allow rapid adaptation to tasks sampled from a distribution. Whilst in simple cases MAML is more data efficient than the Baldwin effect, the Baldwin effect is more general in that it does not require gradients to be backpropagated to the reference parameters or hyperparameters, and permits effectively any number of gradient updates in the inner loop. The Baldwin effect learns strong learning dependent biases, rather than purely genetically accommodating fixed behaviours in a learning independent manner." ] }
1906.08865	2949936448	This paper presents a technique called evolving self-supervised neural networks - neural networks that can teach themselves, intrinsically motivated, without external supervision or reward. The proposed method presents some sort-of paradigm shift, and differs greatly from both traditional gradient-based learning and evolutionary algorithms in that it combines the metaphor of evolution and learning, more specifically self-learning, together, rather than treating these phenomena alternatively. I simulate a multi-agent system in which neural networks are used to control autonomous foraging agents with little domain knowledge. Experimental results show that only evolved self-supervised agents can demonstrate some sort of intelligent behaviour, but not evolution or self-learning alone. Indications for future work on evolving intelligence are also presented.	Generally learning in neural networks can be thought of as part of neural plasticity. There have been some other ideas, like evolving local learning rules to update the weights @cite_13 , evolution of neuromodulation which facilitates the information transfer between neurons in hopes of creating meta-learning @cite_14 . Please refer to @cite_23 for more recent studies on evolving plastic neural networks. In short, most of the work use and neural networks in single-agent environment, having no mutual interaction as they solve their own problems, having no effect on other's performance.	{ "cite_N": [ "@cite_14", "@cite_13", "@cite_23" ], "mid": [ "2161563886", "1598377843", "2605353576" ], "abstract": [ "This paper presents a computational theory on the roles of the ascending neuromodulatory systems from the viewpoint that they mediate the global signals that regulate the distributed learning mechanisms in the brain. Based on the review of experimental data and theoretical models, it is proposed that dopamine signals the error in reward prediction, serotonin controls the time scale of reward prediction, noradrenaline controls the randomness in action selection, and acetylcholine controls the speed of memory update. The possible interactions between those neuromodulators and the environment are predicted on the basis of computational theory of met alearning.", "This paper presents a new approach to neural modeling based on the idea of using an automated method to optimize the parameters of a synaptic learning rule The synaptic modi cation rule is considered as a parametric function This function has local inputs and is the same in many neurons We can use standard optimization methods to select appropriate parameters for a given type of task We also present a theoretical analysis permitting to study the generalization property of such parametric learning rules By generalization we mean the possibility for the learning rule to learn to solve new tasks Experiments were performed on three types of problems a biologically inspired circuit for conditioning in Aplysia Boolean functions linearly separable as well as non linearly separable and classi cation tasks The neural network architecture as well as the form and initial parameter values of the synaptic learning function can be designed using a priori knowledge", "Abstract Biological neural networks are systems of extraordinary computational capabilities shaped by evolution, development, and lifelong learning. The interplay of these elements leads to the emergence of biological intelligence. Inspired by such intricate natural phenomena, Evolved Plastic Artificial Neural Networks (EPANNs) employ simulated evolution in-silico to breed plastic neural networks with the aim to autonomously design and create learning systems. EPANN experiments evolve networks that include both innate properties and the ability to change and learn in response to experiences in different environments and problem domains. EPANNs’ aims include autonomously creating learning systems, bootstrapping learning from scratch, recovering performance in unseen conditions, testing the computational advantages of particular neural components, and deriving hypotheses on the emergence of biological learning. Thus, EPANNs may include a large variety of different neuron types and dynamics, network architectures, plasticity rules, and other factors. While EPANNs have seen considerable progress over the last two decades, current scientific and technological advances in artificial neural networks are setting the conditions for radically new approaches and results. Exploiting the increased availability of computational resources and of simulation environments, the often challenging task of hand-designing learning neural networks could be replaced by more autonomous and creative processes. This paper brings together a variety of inspiring ideas that define the field of EPANNs. The main methods and results are reviewed. Finally, new opportunities and possible developments are presented." ] }
1906.08865	2949936448	This paper presents a technique called evolving self-supervised neural networks - neural networks that can teach themselves, intrinsically motivated, without external supervision or reward. The proposed method presents some sort-of paradigm shift, and differs greatly from both traditional gradient-based learning and evolutionary algorithms in that it combines the metaphor of evolution and learning, more specifically self-learning, together, rather than treating these phenomena alternatively. I simulate a multi-agent system in which neural networks are used to control autonomous foraging agents with little domain knowledge. Experimental results show that only evolved self-supervised agents can demonstrate some sort of intelligent behaviour, but not evolution or self-learning alone. Indications for future work on evolving intelligence are also presented.	In this paper, I adopt a different method called evolving self-supervised neural networks (SSNNs) (which was presented in @cite_18 ). This technique differs from what has been presented so far in this section in that a SSNN consists of two separate network modules, namely action module and reinforcement module. The agent learns as the reinforcement module produces a signal to update the weights of the action module. This happens inside the agent whenever the agent moves and senses the environment, in need of no external , thus intrinsically motivated. This technique also differs greatly from traditional supervised learning in which a learning machine is provided with labels, and from reinforcement learning in which a reward function is set by the system engineer.	{ "cite_N": [ "@cite_18" ], "mid": [ "2947728330" ], "abstract": [ "The so-called Baldwin Effect generally says how learning, as a form of ontogenetic adaptation, can influence the process of phylogenetic adaptation, or evolution. This idea has also been taken into computation in which evolution and learning are used as computational metaphors, including evolving neural networks. This paper presents a technique called evolving self-taught neural networks - neural networks that can teach themselves without external supervision or reward. The self-taught neural network is intrinsically motivated. Moreover, the self-taught neural network is the product of the interplay between evolution and learning. We simulate a multi-agent system in which neural networks are used to control autonomous agents. These agents have to forage for resources and compete for their own survival. Experimental results show that the interaction between evolution and the ability to teach oneself in self-taught neural networks outperform evolution and self-teaching alone. More specifically, the emergence of an intelligent foraging strategy is also demonstrated through that interaction. Indications for future work on evolving neural networks are also presented." ] }
1906.08865	2949936448	This paper presents a technique called evolving self-supervised neural networks - neural networks that can teach themselves, intrinsically motivated, without external supervision or reward. The proposed method presents some sort-of paradigm shift, and differs greatly from both traditional gradient-based learning and evolutionary algorithms in that it combines the metaphor of evolution and learning, more specifically self-learning, together, rather than treating these phenomena alternatively. I simulate a multi-agent system in which neural networks are used to control autonomous foraging agents with little domain knowledge. Experimental results show that only evolved self-supervised agents can demonstrate some sort of intelligent behaviour, but not evolution or self-learning alone. Indications for future work on evolving intelligence are also presented.	Unlike previous work, I simulate a situated multi-agent system. Autonomous multi-agent learning has been stated as a step toward AGI @cite_0 , @cite_25 . here means the intelligent behaviour of an agent is coupled with its environment: the agent controls its own sensory inputs and actions in the environment. Please refer to @cite_10 for a review of situated and embodied approach to cognitive science and AI -- an inevitable part of general intelligence, including that of humans.	{ "cite_N": [ "@cite_0", "@cite_10", "@cite_25" ], "mid": [ "1923871101", "2133196698", "2623431351" ], "abstract": [ "One goal of Artificial Intelligence is to enable the creation of robust, fully autonomous agents that can coexist with us in the real world. Such agents will need to be able to learn, both in order to correct and circumvent their inevitable imperfections, and to keep up with a dynamically changing world. They will also need to be able to interact with one another, whether they share common goals, they pursue independent goals, or their goals are in direct conflict. This paper presents current research directions in machine learning, multiagent reasoning, and robotics, and advocates their unification within concrete application domains. Ideally, new theoretical results in each separate area will inform practical implementations while innovations from concrete multiagent applications will drive new theoretical pursuits, and together these synergistic research approaches will lead us towards the goal of fully autonomous agents.", "The nature of cognition is being re-considered. Instead of emphasizing formal operations on abstract symbols, the new approach foregrounds the fact that cognition is, rather, a situated activity, and suggests that thinking beings ought therefore be considered first and foremost as acting beings. The essay reviews recent work in Embodied Cognition, provides a concise guide to its principles, attitudes and goals, and identifies the physical grounding project as its central research focus.", "We explore deep reinforcement learning methods for multi-agent domains. We begin by analyzing the difficulty of traditional algorithms in the multi-agent case: Q-learning is challenged by an inherent non-stationarity of the environment, while policy gradient suffers from a variance that increases as the number of agents grows. We then present an adaptation of actor-critic methods that considers action policies of other agents and is able to successfully learn policies that require complex multi-agent coordination. Additionally, we introduce a training regimen utilizing an ensemble of policies for each agent that leads to more robust multi-agent policies. We show the strength of our approach compared to existing methods in cooperative as well as competitive scenarios, where agent populations are able to discover various physical and informational coordination strategies." ] }
1906.08989	2949408529	Training a deep network policy for robot manipulation is notoriously costly and time consuming as it depends on collecting a significant amount of real world data. To work well in the real world, the policy needs to see many instances of the task, including various object arrangements in the scene as well as variations in object geometry, texture, material, and environmental illumination. In this paper, we propose a method that learns to perform table-top instance grasping of a wide variety of objects while using no real world grasping data, outperforming the baseline using 2.5D shape by 10 . Our method learns 3D point cloud of object, and use that to train a domain-invariant grasping policy. We formulate the learning process as a two-step procedure: 1) Learning a domain-invariant 3D shape representation of objects from about 76K episodes in simulation and about 530 episodes in the real world, where each episode lasts less than a minute and 2) Learning a critic grasping policy in simulation only based on the 3D shape representation from step 1. Our real world data collection in step 1 is both cheaper and faster compared to existing approaches as it only requires taking multiple snapshots of the scene using a RGBD camera. Finally, the learned 3D representation is not specific to grasping, and can potentially be used in other interaction tasks.	Learning to interact with objects is a wide and actively studied field of vision and robotics research. Many approaches are based on using visual features obtained from RGB or RGBD images to identify objects and grasping points @cite_26 @cite_39 @cite_54 @cite_33 @cite_41 @cite_52 @cite_14 . While early approaches focus on studying the problem of grasping based on traditional techniques, such as logistic regression or learning probabilities of grasp success @cite_26 @cite_39 , more recent approaches often rely on deep neural networks to extract more nuanced features from images. This ranges from detecting objects @cite_54 and their pose @cite_33 , to learning grasp types from kinesthetic demonstrations @cite_41 , under gripper pose uncertainty @cite_4 , or following an unsupervised learning scheme @cite_38 . The effectiveness of deep neural networks is unparalleled, however, training the networks requires large-scale labeled datasets to generalize to unseen objects @cite_14 . Other approaches for robotic grasping focus on identifying the grasp affordance of objects @cite_8 @cite_16 or by categorizing them according to their function @cite_15 .	{ "cite_N": [ "@cite_38", "@cite_26", "@cite_14", "@cite_33", "@cite_4", "@cite_8", "@cite_41", "@cite_54", "@cite_52", "@cite_39", "@cite_15", "@cite_16" ], "mid": [ "2963003918", "2041376653", "2201912979", "", "2485911221", "2140667257", "2300618187", "1999156278", "2603178514", "2148039568", "2003511559", "2041108426" ], "abstract": [ "", "We consider the problem of grasping novel objects, specifically objects that are being seen for the first time through vision. Grasping a previously unknown object, one for which a 3-d model is not available, is a challenging problem. Furthermore, even if given a model, one still has to decide where to grasp the object. We present a learning algorithm that neither requires nor tries to build a 3-d model of the object. Given two (or more) images of an object, our algorithm attempts to identify a few points in each image corresponding to good locations at which to grasp the object. This sparse set of points is then triangulated to obtain a 3-d location at which to attempt a grasp. This is in contrast to standard dense stereo, which tries to triangulate every single point in an image (and often fails to return a good 3-d model). Our algorithm for identifying grasp locations from an image is trained by means of supervised learning, using synthetic images for the training set. We demonstrate this approach on two robotic manipulation platforms. Our algorithm successfully grasps a wide variety of objects, such as plates, tape rolls, jugs, cellphones, keys, screwdrivers, staplers, a thick coil of wire, a strangely shaped power horn and others, none of which were seen in the training set. We also apply our method to the task of unloading items from dishwashers.", "Current model free learning-based robot grasping approaches exploit human-labeled datasets for training the models. However, there are two problems with such a methodology: (a) since each object can be grasped in multiple ways, manually labeling grasp locations is not a trivial task; (b) human labeling is biased by semantics. While there have been attempts to train robots using trial-and-error experiments, the amount of data used in such experiments remains substantially low and hence makes the learner prone to over-fitting. In this paper, we take the leap of increasing the available training data to 40 times more than prior work, leading to a dataset size of 50K data points collected over 700 hours of robot grasping attempts. This allows us to train a Convolutional Neural Network (CNN) for the task of predicting grasp locations without severe overfitting. In our formulation, we recast the regression problem to an 18-way binary classification over image patches. We also present a multi-stage learning approach where a CNN trained in one stage is used to collect hard negatives in subsequent stages. Our experiments clearly show the benefit of using large-scale datasets (and multi-stage training) for the task of grasping. We also compare to several baselines and show state-of-the-art performance on generalization to unseen objects for grasping.", "", "This paper presents a new method for parallel-jaw grasping of isolated objects from depth images, under large gripper pose uncertainty. Whilst most approaches aim to predict the single best grasp pose from an image, our method first predicts a score for every possible grasp pose, which we denote the grasp function. With this, it is possible to achieve grasping robust to the gripper's pose uncertainty, by smoothing the grasp function with the pose uncertainty function. Therefore, if the single best pose is adjacent to a region of poor grasp quality, that pose will no longer be chosen, and instead a pose will be chosen which is surrounded by a region of high grasp quality. To learn this function, we train a Convolutional Neural Network which takes as input a single depth image of an object, and outputs a score for each grasp pose across the image. Training data for this is generated by use of physics simulation and depth image simulation with 3D object meshes, to enable acquisition of sufficient data without requiring exhaustive real-world experiments. We evaluate with both synthetic and real experiments, and show that the learned grasp score is more robust to gripper pose uncertainty than when this uncertainty is not accounted for.", "We present an example-based planning framework to generate semantic grasps, stable grasps that are functionally suitable for specific object manipulation tasks. We propose to use partial object geometry, tactile contacts, and hand kinematic data as proxies to encode task-related constraints, which we call semantic constraints. We introduce a semantic affordance map, which relates local geometry to a set of predefined semantic grasps that are appropriate to different tasks. Using this map, the pose of a robot hand with respect to the object can be estimated so that the hand is adjusted to achieve the ideal approach direction required by a particular task. A grasp planner is then used to search along this approach direction and generate a set of final grasps which have appropriate stability, tactile contacts, and hand kinematics. We show experiments planning semantic grasps on everyday objects and applying these grasps with a physical robot.", "This paper presents a method for one-shot learning of dexterous grasps and grasp generation for novel objects. A model of each grasp type is learned from a single kinesthetic demonstration and several types are taught. These models are used to select and generate grasps for unfamiliar objects. Both the learning and generation stages use an incomplete point cloud from a depth camera, so no prior model of an object shape is used. The learned model is a product of experts, in which experts are of two types. The first type is a contact model and is a density over the pose of a single hand link relative to the local object surface. The second type is the hand-configuration model and is a density over the whole-hand configuration. Grasp generation for an unfamiliar object optimizes the product of these two model types, generating thousands of grasp candidates in under 30 seconds. The method is robust to incomplete data at both training and testing stages. When several grasp types are considered the method selects the highest-likelihood grasp across all the types. In an experiment, the training set consisted of five different grasps and the test set of 45 previously unseen objects. The success rate of the first-choice grasp is 84.4 or 77.7 if seven views or a single view of the test object are taken, respectively.", "We consider the problem of detecting robotic grasps in an RGB-D view of a scene containing objects. In this work, we apply a deep learning approach to solve this problem, which avoids time-consuming hand-design of features. This presents two main challenges. First, we need to evaluate a huge number of candidate grasps. In order to make detection fast and robust, we present a two-step cascaded system with two deep networks, where the top detections from the first are re-evaluated by the second. The first network has fewer features, is faster to run, and can effectively prune out unlikely candidate grasps. The second, with more features, is slower but has to run only on the top few detections. Second, we need to handle multimodal inputs effectively, for which we present a method that applies structured regularization on the weights based on multimodal group regularization. We show that our method improves performance on an RGBD robotic grasping dataset, and can be used to successfully execute grasps on two different robotic platforms.", "Robotic grasping has attracted considerable interest, but it still remains a challenging task. The data-driven approach is a promising solution to the robotic grasping problem; this approach leverages a grasp dataset and generalizes grasps for various objects. However, these methods often depend on the quality of the given datasets, which are not trivial to obtain with sufficient quality. Although reinforcement learning approaches have been recently used to achieve autonomous collection of grasp datasets, the existing algorithms are often limited to specific grasp types. In this paper, we present a framework for hierarchical reinforcement learning of grasping policies. In our framework, the lower-level hierarchy learns multiple grasp types, and the upper-level hierarchy learns a policy to select from the learned grasp types according to a point cloud of a new object. Through experiments, we validate that our approach learns grasping by constructing the grasp dataset autonomously. The experimental results show that our approach learns multiple grasping policies and generalizes the learned grasps by using local point cloud information.", "One of the basic skills for a robot autonomous grasping is to select the appropriate grasping point for an object. Several recent works have shown that it is possible to learn grasping points from different types of features extracted from a single image or from more complex 3D reconstructions. In the context of learning through experience, this is very convenient, since it does not require a full reconstruction of the object and implicitly incorporates kinematic constraints as the hand morphology. These learning strategies usually require a large set of labeled examples which can be expensive to obtain. In this paper, we address the problem of actively learning good grasping points to reduce the number of examples needed by the robot. The proposed algorithm computes the probability of successfully grasping an object at a given location represented by a feature vector. By autonomously exploring different feature values on different objects, the systems learn where to grasp each of the objects. The algorithm combines beta-binomial distributions and a non-parametric kernel approach to provide the full distribution for the probability of grasping. This information allows to perform an active exploration that efficiently learns good grasping points even among different objects. We tested our algorithm using a real humanoid robot that acquired the examples by experimenting directly on the objects and, therefore, it deals better with complex (anthropomorphic) hand-object interactions whose results are difficult to model, or predict. The results show a smooth generalization even in the presence of very few data as is often the case in learning through experience.", "Abstract The problem of finding stable grasps has been widely studied in robotics. However, in many applications the resulting grasps should not only be stable but also applicable for a particular task. Task-specific grasps are closely linked to object categories so that objects in a same category can be often used to perform the same task. This paper presents a probabilistic approach for task-specific stable grasping of objects with shape variations inside the category. An optimal grasp is found as a grasp that is maximally likely to be task compatible and stable taking into account shape uncertainty in a probabilistic context. The method requires only partial models of new objects for grasp generation and only few models and example grasps are used during the training stage. The experiments show that the approach can use multiple models to generalize to new objects in that it outperforms grasping based on the closest model. The method is shown to generate stable grasps for new objects belonging to the same class as well as for similar in shape objects of different categories.", "Autonomous manipulation in unstructured environments will enable a large variety of exciting and important applications. Despite its promise, autonomous manipulation remains largely unsolved. Even the most rudimentary manipulation task--such as removing objects from a pile--remains challenging for robots. We identify three major challenges that must be addressed to enable autonomous manipulation: object segmentation, action selection, and motion generation. These challenges become more pronounced when unknown man-made or natural objects are cluttered together in a pile. We present a system capable of manipulating unknown objects in such an environment. Our robot is tasked with clearing a table by removing objects from a pile and placing them into a bin. To that end, we address the three aforementioned challenges. Our robot perceives the environment with an RGB-D sensor, segmenting the pile into object hypotheses using non-parametric surface models. Our system then computes the affordances of each object, and selects the best affordance and its associated action to execute. Finally, our robot instantiates the proper compliant motion primitive to safely execute the desired action. For efficient and reliable action selection, we developed a framework for supervised learning of manipulation expertise. To verify the performance of our system, we conducted dozens of trials and report on several hours of experiments involving more than 1,500 interactions. The results show that our learning-based approach for pile manipulation outperforms a common sense heuristic as well as a random strategy, and is on par with human action selection." ] }
1906.08989	2949408529	Training a deep network policy for robot manipulation is notoriously costly and time consuming as it depends on collecting a significant amount of real world data. To work well in the real world, the policy needs to see many instances of the task, including various object arrangements in the scene as well as variations in object geometry, texture, material, and environmental illumination. In this paper, we propose a method that learns to perform table-top instance grasping of a wide variety of objects while using no real world grasping data, outperforming the baseline using 2.5D shape by 10 . Our method learns 3D point cloud of object, and use that to train a domain-invariant grasping policy. We formulate the learning process as a two-step procedure: 1) Learning a domain-invariant 3D shape representation of objects from about 76K episodes in simulation and about 530 episodes in the real world, where each episode lasts less than a minute and 2) Learning a critic grasping policy in simulation only based on the 3D shape representation from step 1. Our real world data collection in step 1 is both cheaper and faster compared to existing approaches as it only requires taking multiple snapshots of the scene using a RGBD camera. Finally, the learned 3D representation is not specific to grasping, and can potentially be used in other interaction tasks.	Another line of research focuses on reconstructing objects and scenes as 3D triangle meshes, which in turn can be used to enable more informed robotic behavior @cite_53 @cite_30 . Reconstructing objects from incomplete scans is challenging. Various approaches exist to reconstruct the geometry of objects, while considering the many facets of the problem @cite_35 @cite_37 @cite_56 @cite_29 @cite_18 @cite_19 . Very recently, Henderson and Ferrari @cite_42 introduced a network architecture to generate 3D meshes, while only providing single image supervision. At a higher level, our method is related to recent work on deep 3D representation learning for robot-object interaction @cite_23 @cite_10 @cite_5 . Moreover, to facilitate the learning of 3D shape representations and the grasping of objects, recent efforts concentrate on curating large shape and grasp repositories @cite_17 @cite_1 @cite_31 .	{ "cite_N": [ "@cite_30", "@cite_35", "@cite_37", "@cite_18", "@cite_53", "@cite_29", "@cite_42", "@cite_1", "@cite_56", "@cite_19", "@cite_23", "@cite_5", "@cite_31", "@cite_10", "@cite_17" ], "mid": [ "2570319460", "1987648924", "2427751284", "", "1706124764", "2963640720", "2963338719", "", "2885163231", "2557465155", "2808492412", "2952789668", "2600030077", "2962722175", "2190691619" ], "abstract": [ "In this work, we present a part-based grasp planning approach that is capable of generating grasps that are applicable to multiple familiar objects. We show how object models can be decomposed according to their shape and local volumetric information. The resulting object parts are labeled with semantic information and used for generating robotic grasping information. We investigate how the transfer of such grasping information to familiar objects can be achieved and how the transferability of grasps can be measured. We show that the grasp transferability measure provides valuable information about how successful planned grasps can be applied to novel object instances of the same object category. We evaluate the approach in simulation, by applying it to multiple object categories and determine how successful the planned grasps can be transferred to novel, but familiar objects. In addition, we present a use case on the humanoid robot ARMAR-III.", "We present a system for accurate real-time mapping of complex and arbitrary indoor scenes in variable lighting conditions, using only a moving low-cost depth camera and commodity graphics hardware. We fuse all of the depth data streamed from a Kinect sensor into a single global implicit surface model of the observed scene in real-time. The current sensor pose is simultaneously obtained by tracking the live depth frame relative to the global model using a coarse-to-fine iterative closest point (ICP) algorithm, which uses all of the observed depth data available. We demonstrate the advantages of tracking against the growing full surface model compared with frame-to-frame tracking, obtaining tracking and mapping results in constant time within room sized scenes with limited drift and high accuracy. We also show both qualitative and quantitative results relating to various aspects of our tracking and mapping system. Modelling of natural scenes, in real-time with only commodity sensor and GPU hardware, promises an exciting step forward in augmented reality (AR), in particular, it allows dense surfaces to be reconstructed in real-time, with a level of detail and robustness beyond any solution yet presented using passive computer vision.", "This paper proposes a field model for repairing 3D shapes constructed from multi-view RGB data. Specifically, we represent a 3D shape in a Markov random field (MRF) in which the geometric information is encoded by random binary variables and the appearance information is retrieved from a set of RGB images captured at multiple viewpoints. The local priors in the MRF model capture the local structures of object shapes and are learnt from 3D shape templates using a convolutional deep belief network. Repairing a 3D shape is formulated as the maximum a posteriori (MAP) estimation in the corresponding MRF. Variational mean field approximation technique is adopted for the MAP estimation. The proposed method was evaluated on both artificial data and real data obtained from reconstruction of practical scenes. Experimental results have shown the robustness and efficiency of the proposed method in repairing noisy and incomplete 3D shapes.", "", "An important challenge in robotics is to achieve robust performance in object grasping and manipulation, dealing with noise and uncertainty. This paper presents an approach for addressing the performance of dexterous grasping under shape uncertainty. In our approach, the uncertainty in object shape is parametrized and incorporated as a constraint into grasp planning. The proposed approach is used to plan feasible hand configurations for realizing planned contacts using different robotic hands. A compliant finger closing scheme is devised by exploiting both the object shape uncertainty and tactile sensing at fingertips. Experimental evaluation demonstrates that our method improves the performance of dexterous grasping under shape uncertainty. We considered object shape uncertainty in grasp planning and control.We proposed a probabilistic model to solve hand inverse kinematics.Our grasp planning approach is hand interchangeable.We presented a compliant uncertainty-aware controller for finger closing during grasp execution.", "We introduce ScanComplete, a novel data-driven approach for taking an incomplete 3D scan of a scene as input and predicting a complete 3D model along with per-voxel semantic labels. The key contribution of our method is its ability to handle large scenes with varying spatial extent, managing the cubic growth in data size as scene size increases. To this end, we devise a fully-convolutional generative 3D CNN model whose filter kernels are invariant to the overall scene size. The model can be trained on scene subvolumes but deployed on arbitrarily large scenes at test time. In addition, we propose a coarse-to-fine inference strategy in order to produce high-resolution output while also leveraging large input context sizes. In an extensive series of experiments, we carefully evaluate different model design choices, considering both deterministic and probabilistic models for completion and semantic inference. Our results show that we outperform other methods not only in the size of the environments handled and processing efficiency, but also with regard to completion quality and semantic segmentation performance by a significant margin.", "", "", "Real-life man-made objects often exhibit strong and easily-identifiable structure, as a direct result of their design or their intended functionality. Structure typically appears in the form of individual parts and their arrangement. Knowing about object structure can be an important cue for object recognition and scene understanding - a key goal for various AR and robotics applications. However, commodity RGB-D sensors used in these scenarios only produce raw, unorganized point clouds, without structural information about the captured scene. Moreover, the generated data is commonly partial and susceptible to artifacts and noise, which makes inferring the structure of scanned objects challenging. In this paper, we organize large shape collections into parameterized shape templates to capture the underlying structure of the objects. The templates allow us to transfer the structural information onto new objects and incomplete scans. We employ a deep neural network that matches the partial scan with one of the shape templates, then match and fit it to complete and detailed models from the collection. This allows us to faithfully label its parts and to guide the reconstruction of the scanned object. We showcase the effectiveness of our method by comparing it to other state-of-the-art approaches.", "This paper focuses on semantic scene completion, a task for producing a complete 3D voxel representation of volumetric occupancy and semantic labels for a scene from a single-view depth map observation. Previous work has considered scene completion and semantic labeling of depth maps separately. However, we observe that these two problems are tightly intertwined. To leverage the coupled nature of these two tasks, we introduce the semantic scene completion network (SSCNet), an end-to-end 3D convolutional network that takes a single depth image as input and simultaneously outputs occupancy and semantic labels for all voxels in the camera view frustum. Our network uses a dilation-based 3D context module to efficiently expand the receptive field and enable 3D context learning. To train our network, we construct SUNCG - a manually created largescale dataset of synthetic 3D scenes with dense volumetric annotations. Our experiments demonstrate that the joint model outperforms methods addressing each task in isolation and outperforms alternative approaches on the semantic scene completion task. The dataset and code is available at http: sscnet.cs.princeton.edu.", "Scene representation—the process of converting visual sensory data into concise descriptions—is a requirement for intelligent behavior. Recent work has shown that neural networks excel at this task when provided with large, labeled datasets. However, removing the reliance on human labeling remains an important open problem. To this end, we introduce the Generative Query Network (GQN), a framework within which machines learn to represent scenes using only their own sensors. The GQN takes as input images of a scene taken from different viewpoints, constructs an internal representation, and uses this representation to predict the appearance of that scene from previously unobserved viewpoints. The GQN demonstrates representation learning without human labels or domain knowledge, paving the way toward machines that autonomously learn to understand the world around them.", "Perceiving accurate 3D object shape is important for robots to interact with the physical world. Current research along this direction has been primarily relying on visual observations. Vision, however useful, has inherent limitations due to occlusions and the 2D-3D ambiguities, especially for perception with a monocular camera. In contrast, touch gets precise local shape information, though its efficiency for reconstructing the entire shape could be low. In this paper, we propose a novel paradigm that efficiently perceives accurate 3D object shape by incorporating visual and tactile observations, as well as prior knowledge of common object shapes learned from large-scale shape repositories. We use vision first, applying neural networks with learned shape priors to predict an object's 3D shape from a single-view color image. We then use tactile sensing to refine the shape; the robot actively touches the object regions where the visual prediction has high uncertainty. Our method efficiently builds the 3D shape of common objects from a color image and a small number of tactile explorations (around 10). Our setup is easy to apply and has potentials to help robots better perform grasping or manipulation tasks on real-world objects.", "To reduce data collection time for deep learning of robust robotic grasp plans, we explore training from a synthetic dataset of 6.7 million point clouds, grasps, and analytic grasp metrics generated from thousands of 3D models from Dex-Net 1.0 in randomized poses on a table. We use the resulting dataset, Dex-Net 2.0, to train a Grasp Quality Convolutional Neural Network (GQ-CNN) model that rapidly predicts the probability of success of grasps from depth images, where grasps are specified as the planar position, angle, and depth of a gripper relative to an RGB-D sensor. Experiments with over 1,000 trials on an ABB YuMi comparing grasp planning methods on singulated objects suggest that a GQ-CNN trained with only synthetic data from Dex-Net 2.0 can be used to plan grasps in 0.8sec with a success rate of 93 on eight known objects with adversarial geometry and is 3x faster than registering point clouds to a precomputed dataset of objects and indexing grasps. The Dex-Net 2.0 grasp planner also has the highest success rate on a dataset of 10 novel rigid objects and achieves 99 precision (one false positive out of 69 grasps classified as robust) on a dataset of 40 novel household objects, some of which are articulated or deformable. Code, datasets, videos, and supplementary material are available at http: berkeleyautomation.github.io dex-net .", "This paper focuses on the problem of learning 6- DOF grasping with a parallel jaw gripper in simulation. Our key idea is constraining and regularizing grasping interaction learning through 3D geometry prediction. We introduce a deep geometry-aware grasping network (DGGN) that decomposes the learning into two steps. First, we learn to build mental geometry-aware representation by reconstructing the scene (i.e., 3D occupancy grid) from RGBD input via generative 3D shape modeling. Second, we learn to predict grasping outcome with its internal geometry-aware representation. The learned outcome prediction model is used to sequentially propose grasping solutions via analysis-by-synthesis optimization. Our contributions are fourfold: (1) To best of our knowledge, we are presenting for the first time a method to learn a 6-DOF grasping net from RGBD input; (2) We build a grasping dataset from demonstrations in virtual reality with rich sensory and interaction annotations. This dataset includes 101 everyday objects spread across 7 categories, additionally, we propose a data augmentation strategy for effective learning; (3) We demonstrate that the learned geometry-aware representation leads to about 10 relative performance improvement over the baseline CNN on grasping objects from our dataset. (4) We further demonstrate that the model generalizes to novel viewpoints and object instances.", "We present ShapeNet: a richly-annotated, large-scale repository of shapes represented by 3D CAD models of objects. ShapeNet contains 3D models from a multitude of semantic categories and organizes them under the WordNet taxonomy. It is a collection of datasets providing many semantic annotations for each 3D model such as consistent rigid alignments, parts and bilateral symmetry planes, physical sizes, keywords, as well as other planned annotations. Annotations are made available through a public web-based interface to enable data visualization of object attributes, promote data-driven geometric analysis, and provide a large-scale quantitative benchmark for research in computer graphics and vision. At the time of this technical report, ShapeNet has indexed more than 3,000,000 models, 220,000 models out of which are classified into 3,135 categories (WordNet synsets). In this report we describe the ShapeNet effort as a whole, provide details for all currently available datasets, and summarize future plans." ] }
1906.08989	2949408529	Training a deep network policy for robot manipulation is notoriously costly and time consuming as it depends on collecting a significant amount of real world data. To work well in the real world, the policy needs to see many instances of the task, including various object arrangements in the scene as well as variations in object geometry, texture, material, and environmental illumination. In this paper, we propose a method that learns to perform table-top instance grasping of a wide variety of objects while using no real world grasping data, outperforming the baseline using 2.5D shape by 10 . Our method learns 3D point cloud of object, and use that to train a domain-invariant grasping policy. We formulate the learning process as a two-step procedure: 1) Learning a domain-invariant 3D shape representation of objects from about 76K episodes in simulation and about 530 episodes in the real world, where each episode lasts less than a minute and 2) Learning a critic grasping policy in simulation only based on the 3D shape representation from step 1. Our real world data collection in step 1 is both cheaper and faster compared to existing approaches as it only requires taking multiple snapshots of the scene using a RGBD camera. Finally, the learned 3D representation is not specific to grasping, and can potentially be used in other interaction tasks.	It has been recognized that point sets can serve as an effective representation to obtain additional information of objects. Several recent approaches learn representations to generate point clouds of shapes @cite_44 @cite_43 @cite_20 . More closely targeted to robotics, @cite_45 and @cite_12 propose fusion networks to extract pixel-wise dense feature embeddings for estimating 3D bounding boxes and 6D object pose respectively. These approaches focus on robotic vision, and unlike them we do not rely on the 6D object pose as a label for training.	{ "cite_N": [ "@cite_44", "@cite_43", "@cite_45", "@cite_20", "@cite_12" ], "mid": [ "2560722161", "2784996692", "2962888833", "2850910281", "2910399693" ], "abstract": [ "Generation of 3D data by deep neural network has been attracting increasing attention in the research community. The majority of extant works resort to regular representations such as volumetric grids or collection of images, however, these representations obscure the natural invariance of 3D shapes under geometric transformations, and also suffer from a number of other issues. In this paper we address the problem of 3D reconstruction from a single image, generating a straight-forward form of output – point cloud coordinates. Along with this problem arises a unique and interesting issue, that the groundtruth shape for an input image may be ambiguous. Driven by this unorthordox output form and the inherent ambiguity in groundtruth, we design architecture, loss function and learning paradigm that are novel and effective. Our final solution is a conditional shape sampler, capable of predicting multiple plausible 3D point clouds from an input image. In experiments not only can our system outperform state-of-the-art methods on single image based 3D reconstruction benchmarks, but it also shows strong performance for 3D shape completion and promising ability in making multiple plausible predictions.", "Three-dimensional geometric data offer an excellent domain for studying representation learning and generative modeling. In this paper, we look at geometric data represented as point clouds. We introduce a deep autoencoder (AE) network with excellent reconstruction quality and generalization ability. The learned representations outperform the state of the art in 3D recognition tasks and enable basic shape editing applications via simple algebraic manipulations, such as semantic part editing, shape analogies and shape interpolation. We also perform a thorough study of different generative models including GANs operating on the raw point clouds, significantly improved GANs trained in the fixed latent space our AEs and, Gaussian mixture models (GMM). Interestingly, GMMs trained in the latent space of our AEs produce samples of the best fidelity and diversity. To perform our quantitative evaluation of generative models, we propose simple measures of fidelity and diversity based on optimally matching between sets point clouds.", "We present PointFusion, a generic 3D object detection method that leverages both image and 3D point cloud information. Unlike existing methods that either use multistage pipelines or hold sensor and dataset-specific assumptions, PointFusion is conceptually simple and application-agnostic. The image data and the raw point cloud data are independently processed by a CNN and a PointNet architecture, respectively. The resulting outputs are then combined by a novel fusion network, which predicts multiple 3D box hypotheses and their confidences, using the input 3D points as spatial anchors. We evaluate PointFusion on two distinctive datasets: the KITTI dataset that features driving scenes captured with a lidar-camera setup, and the SUN-RGBD dataset that captures indoor environments with RGB-D cameras. Our model is the first one that is able to perform better or on-par with the state-of-the-art on these diverse datasets without any dataset-specific model tuning.", "We present multiresolution tree-structured networks to process point clouds for 3D shape understanding and generation tasks. Our network represents a 3D shape as a set of locality-preserving 1D ordered list of points at multiple resolutions. This allows efficient feed-forward processing through 1D convolutions, coarse-to-fine analysis through a multi-grid architecture, and it leads to faster convergence and small memory footprint during training. The proposed tree-structured encoders can be used to classify shapes and outperform existing point-based architectures on shape classification benchmarks, while tree-structured decoders can be used for generating point clouds directly and they outperform existing approaches for image-to-shape inference tasks learned using the ShapeNet dataset. Our model also allows unsupervised learning of point-cloud based shapes by using a variational autoencoder, leading to higher-quality generated shapes.", "A key technical challenge in performing 6D object pose estimation from RGB-D image is to fully leverage the two complementary data sources. Prior works either extract information from the RGB image and depth separately or use costly post-processing steps, limiting their performances in highly cluttered scenes and real-time applications. In this work, we present DenseFusion, a generic framework for estimating 6D pose of a set of known objects from RGB-D images. DenseFusion is a heterogeneous architecture that processes the two data sources individually and uses a novel dense fusion network to extract pixel-wise dense feature embedding, from which the pose is estimated. Furthermore, we integrate an end-to-end iterative pose refinement procedure that further improves the pose estimation while achieving near real-time inference. Our experiments show that our method outperforms state-of-the-art approaches in two datasets, YCB-Video and LineMOD. We also deploy our proposed method to a real robot to grasp and manipulate objects based on the estimated pose." ] }
1906.08989	2949408529	Training a deep network policy for robot manipulation is notoriously costly and time consuming as it depends on collecting a significant amount of real world data. To work well in the real world, the policy needs to see many instances of the task, including various object arrangements in the scene as well as variations in object geometry, texture, material, and environmental illumination. In this paper, we propose a method that learns to perform table-top instance grasping of a wide variety of objects while using no real world grasping data, outperforming the baseline using 2.5D shape by 10 . Our method learns 3D point cloud of object, and use that to train a domain-invariant grasping policy. We formulate the learning process as a two-step procedure: 1) Learning a domain-invariant 3D shape representation of objects from about 76K episodes in simulation and about 530 episodes in the real world, where each episode lasts less than a minute and 2) Learning a critic grasping policy in simulation only based on the 3D shape representation from step 1. Our real world data collection in step 1 is both cheaper and faster compared to existing approaches as it only requires taking multiple snapshots of the scene using a RGBD camera. Finally, the learned 3D representation is not specific to grasping, and can potentially be used in other interaction tasks.	Reconstructing the geometry of an object allows identifying grasping points more precisely and thereby to control the grasp @cite_58 @cite_59 @cite_55 . To this end, @cite_0 and @cite_10 recently proposed to reconstruct shapes by learning geometry-aware representations based on 3D occupancy grids. Grids serve as an efficient representation but they obscure the natural invariance of 3D shapes under geometric transformations and only support to represent shapes at low resolution. Similar to our method these approaches also aim at reconstructing objects or object parts to enable more robust grasping. However, unlike them we focus on the self-supervised reconstruction of full 3D point clouds of objects. Point clouds enable a more robust sim-to-real transfer and thereby facilitate efficient training.	{ "cite_N": [ "@cite_55", "@cite_0", "@cite_59", "@cite_58", "@cite_10" ], "mid": [ "2165603175", "2951316215", "1808974695", "2126496149", "2962722175" ], "abstract": [ "This paper presents work on vision based robotic grasping. The proposed method adopts a learning framework where prototypical grasping points are learnt from several examples and then used on novel objects. For representation purposes, we apply the concept of shape context and for learning we use a supervised learning approach in which the classifier is trained with labelled synthetic images. We evaluate and compare the performance of linear and non-linear classifiers. Our results show that a combination of a descriptor based on shape context with a non-linear classification algorithm leads to a stable detection of grasping points for a variety of objects.", "This work provides an architecture to enable robotic grasp planning via shape completion. Shape completion is accomplished through the use of a 3D convolutional neural network (CNN). The network is trained on our own new open source dataset of over 440,000 3D exemplars captured from varying viewpoints. At runtime, a 2.5D pointcloud captured from a single point of view is fed into the CNN, which fills in the occluded regions of the scene, allowing grasps to be planned and executed on the completed object. Runtime shape completion is very rapid because most of the computational costs of shape completion are borne during offline training. We explore how the quality of completions vary based on several factors. These include whether or not the object being completed existed in the training data and how many object models were used to train the network. We also look at the ability of the network to generalize to novel objects allowing the system to complete previously unseen objects at runtime. Finally, experimentation is done both in simulation and on actual robotic hardware to explore the relationship between completion quality and the utility of the completed mesh model for grasping.", "Simulation is essential for different robotic research fields such as mobile robotics, motion planning and grasp planning. For grasping in particular, there are no software simulation packages, which provide a holistic environment that can deal with the variety of aspects associated with this problem. These aspects include development and testing of new algorithms, modeling of the environments and robots, including the modeling of actuators, sensors and contacts. In this paper, we present a new simulation toolkit for grasping and dexterous manipulation called OpenGRASP addressing those aspects in addition to extensibility, interoperability and public availability. OpenGRASP is based on a modular architecture, that supports the creation and addition of new functionality and the integration of existing and widely-used technologies and standards. In addition, a designated editor has been created for the generation and migration of such models. We demonstrate the current state of OpenGRASP's development and its application in a grasp evaluation environment.", "Collecting grasp data for learning and benchmarking purposes is very expensive. It would be helpful to have a standard database of graspable objects, along with a set of stable grasps for each object, but no such database exists. In this work we show how to automate the construction of a database consisting of several hands, thousands of objects, and hundreds of thousands of grasps. Using this database, we demonstrate a novel grasp planning algorithm that exploits geometric similarity between a 3D model and the objects in the database to synthesize form closure grasps. Our contributions are this algorithm, and the database itself, which we are releasing to the community as a tool for both grasp planning and benchmarking.", "This paper focuses on the problem of learning 6- DOF grasping with a parallel jaw gripper in simulation. Our key idea is constraining and regularizing grasping interaction learning through 3D geometry prediction. We introduce a deep geometry-aware grasping network (DGGN) that decomposes the learning into two steps. First, we learn to build mental geometry-aware representation by reconstructing the scene (i.e., 3D occupancy grid) from RGBD input via generative 3D shape modeling. Second, we learn to predict grasping outcome with its internal geometry-aware representation. The learned outcome prediction model is used to sequentially propose grasping solutions via analysis-by-synthesis optimization. Our contributions are fourfold: (1) To best of our knowledge, we are presenting for the first time a method to learn a 6-DOF grasping net from RGBD input; (2) We build a grasping dataset from demonstrations in virtual reality with rich sensory and interaction annotations. This dataset includes 101 everyday objects spread across 7 categories, additionally, we propose a data augmentation strategy for effective learning; (3) We demonstrate that the learned geometry-aware representation leads to about 10 relative performance improvement over the baseline CNN on grasping objects from our dataset. (4) We further demonstrate that the model generalizes to novel viewpoints and object instances." ] }
1906.08989	2949408529	Training a deep network policy for robot manipulation is notoriously costly and time consuming as it depends on collecting a significant amount of real world data. To work well in the real world, the policy needs to see many instances of the task, including various object arrangements in the scene as well as variations in object geometry, texture, material, and environmental illumination. In this paper, we propose a method that learns to perform table-top instance grasping of a wide variety of objects while using no real world grasping data, outperforming the baseline using 2.5D shape by 10 . Our method learns 3D point cloud of object, and use that to train a domain-invariant grasping policy. We formulate the learning process as a two-step procedure: 1) Learning a domain-invariant 3D shape representation of objects from about 76K episodes in simulation and about 530 episodes in the real world, where each episode lasts less than a minute and 2) Learning a critic grasping policy in simulation only based on the 3D shape representation from step 1. Our real world data collection in step 1 is both cheaper and faster compared to existing approaches as it only requires taking multiple snapshots of the scene using a RGBD camera. Finally, the learned 3D representation is not specific to grasping, and can potentially be used in other interaction tasks.	Methods on sim-to-real transfer aim at training neural networks on simulated data with the goal to operate on real data at inference time. This is also known as domain adaptation, where a model is trained with data points of a source domain to then generalize to a target domain @cite_32 @cite_27 . As simulated data can be generated efficiently with ground-truth labels, a number of methods focus on sim-to-real transfer by enhancing or synthesizing images @cite_26 @cite_6 @cite_28 . More recently, @cite_40 introduced a generative adversarial network pipeline to enhance simulated data to significantly reduce the number of real-world data samples for a grasping task. @cite_2 propose a multi-task domain adaptation framework composed of three grasp prediction towers for instance grasping in cluttered scenes. Finally, @cite_22 introduce a two-stage generator pipeline to translate simulated and real images into a canonical representation to realize sim-to-real transfer. While existing work mostly focuses on introducing architectures to operate on images, we argue that point clouds of objects serve as an effective representation to facilitate sim-to-real transfer.	{ "cite_N": [ "@cite_26", "@cite_22", "@cite_28", "@cite_32", "@cite_6", "@cite_27", "@cite_40", "@cite_2" ], "mid": [ "2041376653", "2905464683", "2290564286", "1982696459", "2962787403", "2590953969", "2962899390", "2962837436" ], "abstract": [ "We consider the problem of grasping novel objects, specifically objects that are being seen for the first time through vision. Grasping a previously unknown object, one for which a 3-d model is not available, is a challenging problem. Furthermore, even if given a model, one still has to decide where to grasp the object. We present a learning algorithm that neither requires nor tries to build a 3-d model of the object. Given two (or more) images of an object, our algorithm attempts to identify a few points in each image corresponding to good locations at which to grasp the object. This sparse set of points is then triangulated to obtain a 3-d location at which to attempt a grasp. This is in contrast to standard dense stereo, which tries to triangulate every single point in an image (and often fails to return a good 3-d model). Our algorithm for identifying grasp locations from an image is trained by means of supervised learning, using synthetic images for the training set. We demonstrate this approach on two robotic manipulation platforms. Our algorithm successfully grasps a wide variety of objects, such as plates, tape rolls, jugs, cellphones, keys, screwdrivers, staplers, a thick coil of wire, a strangely shaped power horn and others, none of which were seen in the training set. We also apply our method to the task of unloading items from dishwashers.", "Real world data, especially in the domain of robotics, is notoriously costly to collect. One way to circumvent this can be to leverage the power of simulation to produce large amounts of labelled data. However, training models on simulated images does not readily transfer to real-world ones. Using domain adaptation methods to cross this \"reality gap\" requires a large amount of unlabelled real-world data, whilst domain randomization alone can waste modeling power. In this paper, we present Randomized-to-Canonical Adaptation Networks (RCANs), a novel approach to crossing the visual reality gap that uses no real-world data. Our method learns to translate randomized rendered images into their equivalent non-randomized, canonical versions. This in turn allows for real images to also be translated into canonical sim images. We demonstrate the effectiveness of this sim-to-real approach by training a vision-based closed-loop grasping reinforcement learning agent in simulation, and then transferring it to the real world to attain 70 zero-shot grasp success on unseen objects, a result that almost doubles the success of learning the same task directly on domain randomization alone. Additionally, by joint finetuning in the real-world with only 5,000 real-world grasps, our method achieves 91 , attaining comparable performance to a state-of-the-art system trained with 580,000 real-world grasps, resulting in a reduction of real-world data by more than 99 .", "This paper considers the problem of grasp pose detection in point clouds. We follow a general algorithmic structure that first generates a large set of 6-DOF grasp candidates and then classifies each of them as a good or a bad grasp. Our focus in this paper is on improving the second step by using depth sensor scans from large online datasets to train a convolutional neural network. We propose two new representations of grasp candidates, and we quantify the effect of using prior knowledge of two forms: instance or category knowledge of the object to be grasped, and pretraining the network on simulated depth data obtained from idealized CAD models. Our analysis shows that a more informative grasp candidate representation as well as pretraining and prior knowledge significantly improve grasp detection. We evaluate our approach on a Baxter Research Robot and demonstrate an average grasp success rate of 93 in dense clutter. This is a 20 improvement compared to our prior work.", "In pattern recognition and computer vision, one is often faced with scenarios where the training data used to learn a model have different distribution from the data on which the model is applied. Regardless of the cause, any distributional change that occurs after learning a classifier can degrade its performance at test time. Domain adaptation tries to mitigate this degradation. In this article, we provide a survey of domain adaptation methods for visual recognition. We discuss the merits and drawbacks of existing domain adaptation approaches and identify promising avenues for research in this rapidly evolving field.", "", "The aim of this paper is to give an overview of domain adaptation and transfer learning with a specific view on visual applications. After a general motivation, we first position domain adaptation in the larger transfer learning problem. Second, we try to address and analyze briefly the state-of-the-art methods for different types of scenarios, first describing the historical shallow methods, addressing both the homogeneous and the heterogeneous domain adaptation methods. Third, we discuss the effect of the success of deep convolutional architectures which led to new type of domain adaptation methods that integrate the adaptation within the deep architecture. Fourth, we overview the methods that go beyond image categorization, such as object detection or image segmentation, video analyses or learning visual attributes. Finally, we conclude the paper with a section where we relate domain adaptation to other machine learning solutions.", "Instrumenting and collecting annotated visual grasping datasets to train modern machine learning algorithms can be extremely time-consuming and expensive. An appealing alternative is to use off-the-shelf simulators to render synthetic data for which ground-truth annotations are generated automatically. Unfortunately, models trained purely on simulated data often fail to generalize to the real world. We study how randomized simulated environments and domain adaptation methods can be extended to train a grasping system to grasp novel objects from raw monocular RGB images. We extensively evaluate our approaches with a total of more than 25,000 physical test grasps, studying a range of simulation conditions and domain adaptation methods, including a novel extension of pixel-level domain adaptation that we term the GraspGAN. We show that, by using synthetic data and domain adaptation, we are able to reduce the number of real-world samples needed to achieve a given level of performance by up to 50 times, using only randomly generated simulated objects. We also show that by using only unlabeled real-world data and our GraspGAN methodology, we obtain real-world grasping performance without any real-world labels that is similar to that achieved with 939,777 labeled real-world samples.", "Learning-based approaches to robotic manipulation are limited by the scalability of data collection and accessibility of labels. In this paper, we present a multi-task domain adaptation framework for instance grasping in cluttered scenes by utilizing simulated robot experiments. Our neural network takes monocular RGB images and the instance segmentation mask of a specified target object as inputs, and predicts the probability of successfully grasping the specified object for each candidate motor command. The proposed transfer learning framework trains a model for instance grasping in simulation and uses a domain-adversarial loss to transfer the trained model to real robots using indiscriminate grasping data, which is available both in simulation and the real world. We evaluate our model in real-world robot experiments, comparing it with alternative model architectures as well as an indiscriminate grasping baseline." ] }
1906.08899	2952820735	We study the supervised learning problem under either of the following two models: (1) Feature vectors @math are @math -dimensional Gaussians and responses are @math for @math an unknown quadratic function; (2) Feature vectors @math are distributed as a mixture of two @math -dimensional centered Gaussians, and @math 's are the corresponding class labels. We use two-layers neural networks with quadratic activations, and compare three different learning regimes: the random features (RF) regime in which we only train the second-layer weights; the neural tangent (NT) regime in which we train a linearization of the neural network around its initialization; the fully trained neural network (NN) regime in which we train all the weights in the network. We prove that, even for the simple quadratic model of point (1), there is a potentially unbounded gap between the prediction risk achieved in these three training regimes, when the number of neurons is smaller than the ambient dimension. When the number of neurons is larger than the number of dimensions, the problem is significantly easier and both NT and NN learning achieve zero risk.	The connection (and differences) between two-layers neural networks and random features models has been the object of several papers since the original work of Rahimi and Recht @cite_29 . An incomplete list of references includes @cite_33 @cite_14 @cite_11 @cite_23 @cite_31 . Our analysis contributes to this line of work by establishing a sharp asymptotic characterization, although in more specific data distributions. Sharp results have recently been proven in @cite_18 , for the special case of random weights @math uniformly distributed over a @math -dimensional sphere. Here we consider the more general case of anisotropic random features with covariance @math . This clarifies a key reason for suboptimality of random features: the data representation is not adapted to the target function @math . We focus on the population limit @math . Complementary results characterizing the variance as a function of @math are given in @cite_4 .	{ "cite_N": [ "@cite_18", "@cite_14", "@cite_4", "@cite_33", "@cite_29", "@cite_23", "@cite_31", "@cite_11" ], "mid": [ "2941057241", "", "2923764619", "2963942108", "", "", "2963013450", "2963534251" ], "abstract": [ "We consider the problem of learning an unknown function @math on the @math -dimensional sphere with respect to the square loss, given i.i.d. samples @math where @math is a feature vector uniformly distributed on the sphere and @math . We study two popular classes of models that can be regarded as linearizations of two-layers neural networks around a random initialization: (RF) The random feature model of Rahimi-Recht; (NT) The neural tangent kernel model of Jacot-Gabriel-Hongler. Both these approaches can also be regarded as randomized approximations of kernel ridge regression (with respect to different kernels), and hence enjoy universal approximation properties when the number of neurons @math diverges, for a fixed dimension @math . We prove that, if both @math and @math are large, the behavior of these models is instead remarkably simpler. If @math , then RF performs no better than linear regression with respect to the raw features @math , and NT performs no better than linear regression with respect to degree-one and two monomials in the @math . More generally, if @math then RF fits at most a degree- @math polynomial in the raw features, and NT fits at most a degree- @math polynomial.", "", "Interpolators---estimators that achieve zero training error---have attracted growing attention in machine learning, mainly because state-of-the art neural networks appear to be models of this type. In this paper, we study minimum @math norm ( ridgeless') interpolation in high-dimensional least squares regression. We consider two different models for the feature distribution: a linear model, where the feature vectors @math are obtained by applying a linear transform to a vector of i.i.d. entries, @math (with @math ); and a nonlinear model, where the feature vectors are obtained by passing the input through a random one-layer neural network, @math (with @math , @math a matrix of i.i.d. entries, and @math an activation function acting componentwise on @math ). We recover---in a precise quantitative way---several phenomena that have been observed in large-scale neural networks and kernel machines, including the double descent' behavior of the prediction risk, and the potential benefits of overparametrization.", "We consider supervised learning problems within the positive-definite kernel framework, such as kernel ridge regression, kernel logistic regression or the support vector machine. With kernels leading to infinite-dimensional feature spaces, a common practi cal limiting difficulty is the necessity of computing the kernel matrix, which most frequently leads to algorithms with running time at least quadratic in the number of observations n, i.e., O(n 2 ). Low-rank approximations of the kernel matrix are often considered as they allow the reduction of running time complexities to O(p 2 n), where p is the rank of the approximation. The practicality of such methods thus depends on the required rank p. In this paper, we show that in the context of kernel ridge regression, for approximations based on a random subset of columns of the original kernel matrix, the rank p may be chosen to be linear in the degrees of freedom associated with the problem, a quantity which is classically used in the statistical analysis of such method s, and is often seen as the implicit number of parameters of non-parametric estimators. This result enables simple algorithms that have sub-quadratic running time complexity, but provably exhibit the same predictive performancethan existing algorithms, for any given problem instance, and not only for worst-case situations.", "", "", "We study the generalization properties of ridge regression with random features in the statistical learning framework. We show for the first time that @math learning bounds can be achieved with only @math random features rather than @math as suggested by previous results. Further, we prove faster learning rates and show that they might require more random features, unless they are sampled according to a possibly problem dependent distribution. Our results shed light on the statistical computational trade-offs in large scale kernelized learning, showing the potential effectiveness of random features in reducing the computational complexity while keeping optimal generalization properties.", "We consider neural networks with a single hidden layer and non-decreasing positively homogeneous activation functions like the rectified linear units. By letting the number of hidden units grow unbounded and using classical non-Euclidean regularization tools on the output weights, they lead to a convex optimization problem and we provide a detailed theoretical analysis of their generalization performance, with a study of both the approximation and the estimation errors. We show in particular that they are adaptive to unknown underlying linear structures, such as the dependence on the projection of the input variables onto a low-dimensional subspace. Moreover, when using sparsity-inducing norms on the input weights, we show that high-dimensional non-linear variable selection may be achieved, without any strong assumption regarding the data and with a total number of variables potentially exponential in the number of observations. However, solving this convex optimization problem in infinite dimensions is only possible if the nonconvex subproblem of addition of a new unit can be solved efficiently. We provide a simple geometric interpretation for our choice of activation functions and describe simple conditions for convex relaxations of the finite-dimensional non-convex subproblem to achieve the same generalization error bounds, even when constant-factor approximations cannot be found. We were not able to find strong enough convex relaxations to obtain provably polynomialtime algorithms and leave open the existence or non-existence of such tractable algorithms with non-exponential sample complexities." ] }
1906.08899	2952820735	We study the supervised learning problem under either of the following two models: (1) Feature vectors @math are @math -dimensional Gaussians and responses are @math for @math an unknown quadratic function; (2) Feature vectors @math are distributed as a mixture of two @math -dimensional centered Gaussians, and @math 's are the corresponding class labels. We use two-layers neural networks with quadratic activations, and compare three different learning regimes: the random features (RF) regime in which we only train the second-layer weights; the neural tangent (NT) regime in which we train a linearization of the neural network around its initialization; the fully trained neural network (NN) regime in which we train all the weights in the network. We prove that, even for the simple quadratic model of point (1), there is a potentially unbounded gap between the prediction risk achieved in these three training regimes, when the number of neurons is smaller than the ambient dimension. When the number of neurons is larger than the number of dimensions, the problem is significantly easier and both NT and NN learning achieve zero risk.	The @math model is much more recent @cite_35 . Several papers show that SGD optimization within the original neural network is well approximated by optimization within the model @math as long as the number of neurons is large compared to a polynomial in the sample size @math @cite_34 @cite_17 @cite_36 @cite_7 . Empirical evidence in the same direction was presented in @cite_8 @cite_32 .	{ "cite_N": [ "@cite_35", "@cite_7", "@cite_8", "@cite_36", "@cite_32", "@cite_34", "@cite_17" ], "mid": [ "", "2900959181", "2912322140", "2899748887", "2942052807", "2894604724", "2899790086" ], "abstract": [ "", "We study the problem of training deep neural networks with Rectified Linear Unit (ReLU) activation function using gradient descent and stochastic gradient descent. In particular, we study the binary classification problem and show that for a broad family of loss functions, with proper random weight initialization, both gradient descent and stochastic gradient descent can find the global minima of the training loss for an over-parameterized deep ReLU network, under mild assumption on the training data. The key idea of our proof is that Gaussian random initialization followed by (stochastic) gradient descent produces a sequence of iterates that stay inside a small perturbation region centering around the initial weights, in which the empirical loss function of deep ReLU networks enjoys nice local curvature properties that ensure the global convergence of (stochastic) gradient descent. Our theoretical results shed light on understanding the optimization for deep learning, and pave the way for studying the optimization dynamics of training modern deep neural networks.", "A longstanding goal in deep learning research has been to precisely characterize training and generalization. However, the often complex loss landscapes of neural networks have made a theory of learning dynamics elusive. In this work, we show that for wide neural networks the learning dynamics simplify considerably and that, in the infinite width limit, they are governed by a linear model obtained from the first-order Taylor expansion of the network around its initial parameters. Furthermore, mirroring the correspondence between wide Bayesian neural networks and Gaussian processes, gradient-based training of wide neural networks with a squared loss produces test set predictions drawn from a Gaussian process with a particular compositional kernel. While these theoretical results are only exact in the infinite width limit, we nevertheless find excellent empirical agreement between the predictions of the original network and those of the linearized version even for finite practically-sized networks. This agreement is robust across different architectures, optimization methods, and loss functions.", "Deep neural networks (DNNs) have demonstrated dominating performance in many fields; since AlexNet, networks used in practice are going wider and deeper. On the theoretical side, a long line of works has been focusing on training neural networks with one hidden layer. The theory of multi-layer networks remains largely unsettled. In this work, we prove why stochastic gradient descent (SGD) can find @math on the training objective of DNNs in @math . We only make two assumptions: the inputs are non-degenerate and the network is over-parameterized. The latter means the network width is sufficiently large: @math in @math , the number of layers and in @math , the number of samples. Our key technique is to derive that, in a sufficiently large neighborhood of the random initialization, the optimization landscape is almost-convex and semi-smooth even with ReLU activations. This implies an equivalence between over-parameterized neural networks and neural tangent kernel (NTK) in the finite (and polynomial) width setting. As concrete examples, starting from randomly initialized weights, we prove that SGD can attain 100 training accuracy in classification tasks, or minimize regression loss in linear convergence speed, with running time polynomial in @math . Our theory applies to the widely-used but non-smooth ReLU activation, and to any smooth and possibly non-convex loss functions. In terms of network architectures, our theory at least applies to fully-connected neural networks, convolutional neural networks (CNN), and residual neural networks (ResNet).", "How well does a classic deep net architecture like AlexNet or VGG19 classify on a standard dataset such as CIFAR-10 when its \"width\" --- namely, number of channels in convolutional layers, and number of nodes in fully-connected internal layers --- is allowed to increase to infinity? Such questions have come to the forefront in the quest to theoretically understand deep learning and its mysteries about optimization and generalization. They also connect deep learning to notions such as Gaussian processes and kernels. A recent paper [, 2018] introduced the Neural Tangent Kernel (NTK) which captures the behavior of fully-connected deep nets in the infinite width limit trained by gradient descent; this object was implicit in some other recent papers. A subsequent paper [, 2019] gave heuristic Monte Carlo methods to estimate the NTK and its extension, Convolutional Neural Tangent Kernel (CNTK) and used this to try to understand the limiting behavior on datasets like CIFAR-10. The current paper gives the first efficient exact algorithm (based upon dynamic programming) for computing CNTK as well as an efficient GPU implementation of this algorithm. This results in a significant new benchmark for performance of a pure kernel-based method on CIFAR-10, being 10 higher than the methods reported in [, 2019], and only 5 lower than the performance of the corresponding finite deep net architecture (once batch normalization etc. are turned off). We give the first non-asymptotic proof showing that a fully-trained sufficiently wide net is indeed equivalent to the kernel regression predictor using NTK. Our experiments also demonstrate that earlier Monte Carlo approximation can degrade the performance significantly, thus highlighting the power of our exact kernel computation, which we have applied even to the full CIFAR-10 dataset and 20-layer nets.", "One of the mysteries in the success of neural networks is randomly initialized first order methods like gradient descent can achieve zero training loss even though the objective function is non-convex and non-smooth. This paper demystifies this surprising phenomenon for two-layer fully connected ReLU activated neural networks. For an @math hidden node shallow neural network with ReLU activation and @math training data, we show as long as @math is large enough and no two inputs are parallel, randomly initialized gradient descent converges to a globally optimal solution at a linear convergence rate for the quadratic loss function. Our analysis relies on the following observation: over-parameterization and random initialization jointly restrict every weight vector to be close to its initialization for all iterations, which allows us to exploit a strong convexity-like property to show that gradient descent converges at a global linear rate to the global optimum. We believe these insights are also useful in analyzing deep models and other first order methods.", "Gradient descent finds a global minimum in training deep neural networks despite the objective function being non-convex. The current paper proves gradient descent achieves zero training loss in polynomial time for a deep over-parameterized neural network with residual connections (ResNet). Our analysis relies on the particular structure of the Gram matrix induced by the neural network architecture. This structure allows us to show the Gram matrix is stable throughout the training process and this stability implies the global optimality of the gradient descent algorithm. We further extend our analysis to deep residual convolutional neural networks and obtain a similar convergence result." ] }
1906.08899	2952820735	We study the supervised learning problem under either of the following two models: (1) Feature vectors @math are @math -dimensional Gaussians and responses are @math for @math an unknown quadratic function; (2) Feature vectors @math are distributed as a mixture of two @math -dimensional centered Gaussians, and @math 's are the corresponding class labels. We use two-layers neural networks with quadratic activations, and compare three different learning regimes: the random features (RF) regime in which we only train the second-layer weights; the neural tangent (NT) regime in which we train a linearization of the neural network around its initialization; the fully trained neural network (NN) regime in which we train all the weights in the network. We prove that, even for the simple quadratic model of point (1), there is a potentially unbounded gap between the prediction risk achieved in these three training regimes, when the number of neurons is smaller than the ambient dimension. When the number of neurons is larger than the number of dimensions, the problem is significantly easier and both NT and NN learning achieve zero risk.	Chizat and Bach @cite_12 clarified that any nonlinear statistical model can be approximated by a linear one in an early () training regime. The basic argument is quite simple. Given a model @math with parameters @math , we can Taylor-expand around a random initialization @math . Setting @math , we get Here the second approximation holds since, for many random initializations, @math because of random cancellations. The resulting model @math is linear, with random features.	{ "cite_N": [ "@cite_12" ], "mid": [ "2904838594" ], "abstract": [ "In a series of recent theoretical works, it has been shown that strongly over-parameterized neural networks trained with gradient-based methods could converge linearly to zero training loss, with their parameters hardly varying. In this note, our goal is to exhibit the simple structure that is behind these results. In a simplified setting, we prove that \"lazy training\" essentially solves a kernel regression. We also show that this behavior is not so much due to over-parameterization than to a choice of scaling, often implicit, that allows to linearize the model around its initialization. These theoretical results complemented with simple numerical experiments make it seem unlikely that \"lazy training\" is behind the many successes of neural networks in high dimensional tasks." ] }
1906.08899	2952820735	We study the supervised learning problem under either of the following two models: (1) Feature vectors @math are @math -dimensional Gaussians and responses are @math for @math an unknown quadratic function; (2) Feature vectors @math are distributed as a mixture of two @math -dimensional centered Gaussians, and @math 's are the corresponding class labels. We use two-layers neural networks with quadratic activations, and compare three different learning regimes: the random features (RF) regime in which we only train the second-layer weights; the neural tangent (NT) regime in which we train a linearization of the neural network around its initialization; the fully trained neural network (NN) regime in which we train all the weights in the network. We prove that, even for the simple quadratic model of point (1), there is a potentially unbounded gap between the prediction risk achieved in these three training regimes, when the number of neurons is smaller than the ambient dimension. When the number of neurons is larger than the number of dimensions, the problem is significantly easier and both NT and NN learning achieve zero risk.	Finally, our analysis of fully trained networks connects to the ample literature on non-convex statistical estimation. For two layers neural networks with quadratic activations, Soltanolkotabi, Javanmard and Lee @cite_15 showed that, as long as the number of neurons satisfies @math there are no spurious local minimizers. Du and Lee @cite_22 showed that the same holds as long as @math where @math is the sample size. Zhong et. al. @cite_10 established local convexity properties around global optima. Further related landscape results include @cite_25 @cite_6 @cite_0 .	{ "cite_N": [ "@cite_22", "@cite_6", "@cite_0", "@cite_15", "@cite_10", "@cite_25" ], "mid": [ "2789756211", "2126418337", "2604130501", "2963417959", "2962767131", "2766371994" ], "abstract": [ "We provide new theoretical insights on why over-parametrization is effective in learning neural networks. For a @math hidden node shallow network with quadratic activation and @math training data points, we show as long as @math , over-parametrization enables local search algorithms to find a optimal solution for general smooth and convex loss functions. Further, despite that the number of parameters may exceed the sample size, using theory of Rademacher complexity, we show with weight decay, the solution also generalizes well if the data is sampled from a regular distribution such as Gaussian. To prove when @math , the loss function has benign landscape properties, we adopt an idea from smoothed analysis, which may have other applications in studying loss surfaces of neural networks.", "Recently, convex solutions to low-rank matrix factorization problems have received increasing attention in machine learning. However, in many applications the data can display other structures beyond simply being low-rank. For example, images and videos present complex spatio-temporal structures, which are largely ignored by current low-rank methods. In this paper we explore a matrix factorization technique suitable for large datasets that captures additional structure in the factors by using a projective tensor norm, which includes classical image regularizers such as total variation and the nuclear norm as particular cases. Although the resulting optimization problem is not convex, we show that under certain conditions on the factors, any local minimizer for the factors yields a global minimizer for their product. Examples in biomedical video segmentation and hyperspectral compressed recovery show the advantages of our approach on high-dimensional datasets.", "In this paper we develop a new framework that captures the common landscape underlying the common non-convex low-rank matrix problems including matrix sensing, matrix completion and robust PCA. In particular, we show for all above problems (including asymmetric cases): 1) all local minima are also globally optimal; 2) no high-order saddle points exists. These results explain why simple algorithms such as stochastic gradient descent have global converge, and efficiently optimize these non-convex objective functions in practice. Our framework connects and simplifies the existing analyses on optimization landscapes for matrix sensing and symmetric matrix completion. The framework naturally leads to new results for asymmetric matrix completion and robust PCA.", "In this paper, we study the problem of learning a shallow artificial neural network that best fits a training data set. We study this problem in the over-parameterized regime where the numbers of observations are fewer than the number of parameters in the model. We show that with the quadratic activations, the optimization landscape of training, such shallow neural networks, has certain favorable characteristics that allow globally optimal models to be found efficiently using a variety of local search heuristics. This result holds for an arbitrary training data of input output pairs. For differentiable activation functions, we also show that gradient descent, when suitably initialized, converges at a linear rate to a globally optimal model. This result focuses on a realizable model where the inputs are chosen i.i.d. from a Gaussian distribution and the labels are generated according to planted weight coefficients.", "", "We consider the problem of learning a one-hidden-layer neural network: we assume the input @math is from Gaussian distribution and the label @math , where @math is a nonnegative vector in @math with @math , @math is a full-rank weight matrix, and @math is a noise vector. We first give an analytic formula for the population risk of the standard squared loss and demonstrate that it implicitly attempts to decompose a sequence of low-rank tensors simultaneously. Inspired by the formula, we design a non-convex objective function @math whose landscape is guaranteed to have the following properties: 1. All local minima of @math are also global minima. 2. All global minima of @math correspond to the ground truth parameters. 3. The value and gradient of @math can be estimated using samples. With these properties, stochastic gradient descent on @math provably converges to the global minimum and learn the ground-truth parameters. We also prove finite sample complexity result and validate the results by simulations." ] }
1906.09081	2952480557	Bipartite networks are a well known strategy to study a variety of phenomena. The commonly used method to deal with this type of network is to project the bipartite data into a unipartite weighted graph and then using a backboning technique to extract only the meaningful edges. Despite the wide availability of different methods both for projection and backboning, we believe that there has been little attention to the effect that the combination of these two processes has on the data and on the resulting network topology. In this paper we study the effect that the possible combinations of projection and backboning techniques have on a bipartite network. We show that the 12 methods group into two clusters producing unipartite networks with very different topologies. We also show that the resulting level of network centralization is highly affected by the combination of projection and backboning applied.	Among the many areas that have used bipartite data, the present paper was inspired by a recent study published in the area of communication research, more precisely in the sub-field of online audience studies @cite_24 . In this context, audience networks have been used for several years @cite_7 to study the general structure of audiences and answer key questions about echo-chambers, the existence of trend toward cyber-balkanization and filter bubbles. Audience networks are networks where nodes represents media sites that are connected if they share at least a part of their audience @cite_24 . While there are several strategies to collect audience behaviour, over the recent years the common practice has been the use of trace data -- digital traces left behind by users browsing the web @cite_24 . This is also due to intrinsic limitation of self-reported data @cite_10 .	{ "cite_N": [ "@cite_24", "@cite_10", "@cite_7" ], "mid": [ "2895991954", "2109790075", "1965598481" ], "abstract": [ "Measures of audience overlap between news sources give us information on the diversity of people’s media diets and the similarity of news outlets in terms of the audiences they share. This provides...", "Many studies of media effects use self-reported news exposure as their key independent variable without establishing its validity. Motivated by anecdotal evidence that people's reports of their own media use can differ considerably from independent assessments, this study examines systematically the accuracy of survey-based self-reports of news exposure. I compare survey estimates to Nielsen estimates, which do not rely on self-reports. Results show severe overreporting of news exposure. Survey estimates of network news exposure follow trends in Nielsen ratings relatively well, but exaggerate exposure by a factor of 3 on average and as much as eightfold for some demographics. It follows that apparent media effects may arise not because of differences in exposure, but because of unknown differences in the accuracy of reporting exposure.", "This article explains and implements a network analytic approach to the study of cross-platform audience behavior. It begins by conceptualizing large-scale patterns of media use in network terms, treating media outlets as nodes and the levels of audience duplication among them as links. Following that, it explains 2 common measures of audience duplication, Absolute Duplication and Primary Duplication, and offers a new measure, Deviation-from-Random Duplication. In doing so, techniques for converting duplication data into network data are discussed. This approach is then applied to analyze patterns of audience fragmentation, media publics, and audience polarization using data from Nielsen's TV Internet Convergence Panel. The findings show the value of using a network approach, by contributing to an alternative understanding of these patterns. Economic and policy implications are discussed, as well as broader reflections on the use of network analysis in the study of audience behavior." ] }
1906.08856	2952632952	Learning long-term dependencies still remains difficult for recurrent neural networks (RNNs) despite their success in sequence modeling recently. In this paper, we propose a novel gated RNN structure, which contains only one gate. Hidden states in the proposed grouped distributor unit (GDU) are partitioned into groups. For each group, the proportion of memory to be overwritten in each state transition is limited to a constant and is adaptively distributed to each group member. In other word, every separate group has a fixed overall update rate, yet all units are allowed to have different paces. Information is therefore forced to be latched in a flexible way, which helps the model to capture long-term dependencies in data. Besides having a simpler structure, GDU is demonstrated experimentally to outperform LSTM and GRU on tasks including both pathological problems and natural data set.	An RNN is able to encode sequences of arbitrary length into a fixed-length representation by folding a new observation @math into its hidden state @math using a transition operator @math at each time step @math : Simple recurrent networks (SRN, @cite_22 ), known as one of the earliest variants, make @math as the composition of an element-wise nonlinearity with an affine transformation of both @math and @math : W _s x _t + U _s s _ t-1 + b _s where @math is the input-to-state weight matrix, @math is the state-to-state recurrent weight matrix, @math is the bias and @math is the nonlinear activation function. For the convenience of the following descriptions, we denote this kind of operators as @math , and a subscript can be added to distinguish different network components. Thus in SRN, @math .	{ "cite_N": [ "@cite_22" ], "mid": [ "2110485445" ], "abstract": [ "Time underlies many interesting human behaviors. Thus, the question of how to represent time in connectionist models is very important. One approach is to represent time implicitly by its effects on processing rather than explicitly (as in a spatial representation). The current report develops a proposal along these lines first described by Jordan (1986) which involves the use of recurrent links in order to provide networks with a dynamic memory. In this approach, hidden unit patterns are fed back to themselves; the internal representations which develop thus reflect task demands in the context of prior internal states. A set of simulations is reported which range from relatively simple problems (temporal version of XOR) to discovering syntactic semantic features for words. The networks are able to learn interesting internal representations which incorporate task demands with memory demands; indeed, in this approach the notion of memory is inextricably bound up with task processing. These representations reveal a rich structure, which allows them to be highly context-dependent while also expressing generalizations across classes of items. These representations suggest a method for representing lexical categories and the type token distinction." ] }
1906.08856	2952632952	Learning long-term dependencies still remains difficult for recurrent neural networks (RNNs) despite their success in sequence modeling recently. In this paper, we propose a novel gated RNN structure, which contains only one gate. Hidden states in the proposed grouped distributor unit (GDU) are partitioned into groups. For each group, the proportion of memory to be overwritten in each state transition is limited to a constant and is adaptively distributed to each group member. In other word, every separate group has a fixed overall update rate, yet all units are allowed to have different paces. Information is therefore forced to be latched in a flexible way, which helps the model to capture long-term dependencies in data. Besides having a simpler structure, GDU is demonstrated experimentally to outperform LSTM and GRU on tasks including both pathological problems and natural data set.	In SRN, @math is given by @math . As a result, if the derivative of the nonlinear function is bounded in SRN, namely, @math , s.t. @math , it will be for @math , where @math is the largest singular value of the recurrent weight matrix @math , for @math to vanish (as @math ) @cite_8 .	{ "cite_N": [ "@cite_8" ], "mid": [ "1815076433" ], "abstract": [ "There are two widely known issues with properly training recurrent neural networks, the vanishing and the exploding gradient problems detailed in (1994). In this paper we attempt to improve the understanding of the underlying issues by exploring these problems from an analytical, a geometric and a dynamical systems perspective. Our analysis is used to justify a simple yet effective solution. We propose a gradient norm clipping strategy to deal with exploding gradients and a soft constraint for the vanishing gradients problem. We validate empirically our hypothesis and proposed solutions in the experimental section." ] }
1906.08856	2952632952	Learning long-term dependencies still remains difficult for recurrent neural networks (RNNs) despite their success in sequence modeling recently. In this paper, we propose a novel gated RNN structure, which contains only one gate. Hidden states in the proposed grouped distributor unit (GDU) are partitioned into groups. For each group, the proportion of memory to be overwritten in each state transition is limited to a constant and is adaptively distributed to each group member. In other word, every separate group has a fixed overall update rate, yet all units are allowed to have different paces. Information is therefore forced to be latched in a flexible way, which helps the model to capture long-term dependencies in data. Besides having a simpler structure, GDU is demonstrated experimentally to outperform LSTM and GRU on tasks including both pathological problems and natural data set.	@cite_11 proposed a similar architecture with gating units called gated recurrent unit (GRU). Different from LSTM, GRU exposes all its states to the output and use a linear interpolation between the previous state @math and the candidate state @math :	{ "cite_N": [ "@cite_11" ], "mid": [ "2950635152" ], "abstract": [ "In this paper, we propose a novel neural network model called RNN Encoder-Decoder that consists of two recurrent neural networks (RNN). One RNN encodes a sequence of symbols into a fixed-length vector representation, and the other decodes the representation into another sequence of symbols. The encoder and decoder of the proposed model are jointly trained to maximize the conditional probability of a target sequence given a source sequence. The performance of a statistical machine translation system is empirically found to improve by using the conditional probabilities of phrase pairs computed by the RNN Encoder-Decoder as an additional feature in the existing log-linear model. Qualitatively, we show that the proposed model learns a semantically and syntactically meaningful representation of linguistic phrases." ] }
1906.08856	2952632952	Learning long-term dependencies still remains difficult for recurrent neural networks (RNNs) despite their success in sequence modeling recently. In this paper, we propose a novel gated RNN structure, which contains only one gate. Hidden states in the proposed grouped distributor unit (GDU) are partitioned into groups. For each group, the proportion of memory to be overwritten in each state transition is limited to a constant and is adaptively distributed to each group member. In other word, every separate group has a fixed overall update rate, yet all units are allowed to have different paces. Information is therefore forced to be latched in a flexible way, which helps the model to capture long-term dependencies in data. Besides having a simpler structure, GDU is demonstrated experimentally to outperform LSTM and GRU on tasks including both pathological problems and natural data set.	Previous work has clearly indicated the advantages of the gating units over the more traditional recurrent units @cite_5 . Both LSTM and GRU perform well in tasks that require capturing long-term dependencies. However, the choice of these two structures may depend heavily on the dataset and corresponding task.	{ "cite_N": [ "@cite_5" ], "mid": [ "1924770834" ], "abstract": [ "In this paper we compare different types of recurrent units in recurrent neural networks (RNNs). Especially, we focus on more sophisticated units that implement a gating mechanism, such as a long short-term memory (LSTM) unit and a recently proposed gated recurrent unit (GRU). We evaluate these recurrent units on the tasks of polyphonic music modeling and speech signal modeling. Our experiments revealed that these advanced recurrent units are indeed better than more traditional recurrent units such as tanh units. Also, we found GRU to be comparable to LSTM." ] }
1906.08856	2952632952	Learning long-term dependencies still remains difficult for recurrent neural networks (RNNs) despite their success in sequence modeling recently. In this paper, we propose a novel gated RNN structure, which contains only one gate. Hidden states in the proposed grouped distributor unit (GDU) are partitioned into groups. For each group, the proportion of memory to be overwritten in each state transition is limited to a constant and is adaptively distributed to each group member. In other word, every separate group has a fixed overall update rate, yet all units are allowed to have different paces. Information is therefore forced to be latched in a flexible way, which helps the model to capture long-term dependencies in data. Besides having a simpler structure, GDU is demonstrated experimentally to outperform LSTM and GRU on tasks including both pathological problems and natural data set.	Simplifying GAST has drawn interest of researchers recently. GRU itself reduces the gate units to @math compared to LSTM which has @math gate units by coupling forget gate and input gate into one update gate, namely making the gate operator @math equals to @math . In this paper we denote this kind of GAST as cGAST, with the prefix c short for . Based on GRU, the Minimal Gated Unit (MGU, @cite_10 ) reduced the gate number further to only 1 by letting @math without losing GRU's accuracy benefits. The Update Gate RNN (UGRNN, @cite_3 ) entirely removed @math operator. However, none of these models has shown superiority over LSTM and GRU on long-term tasks with single-layer hidden states.	{ "cite_N": [ "@cite_10", "@cite_3" ], "mid": [ "2319453305", "2557270725" ], "abstract": [ "Recurrent neural networks (RNN) have been very successful in handling sequence data. However, understanding RNN and finding the best practices for RNN learning is a difficult task, partly because there are many competing and complex hidden units, such as the long short-term memory (LSTM) and the gated recurrent unit (GRU). We propose a gated unit for RNN, named as minimal gated unit (MGU), since it only contains one gate, which is a minimal design among all gated hidden units. The design of MGU benefits from evaluation results on LSTM and GRU in the literature. Experiments on various sequence data show that MGU has comparable accuracy with GRU, but has a simpler structure, fewer parameters, and faster training. Hence, MGU is suitable in RNN's applications. Its simple architecture also means that it is easier to evaluate and tune, and in principle it is easier to study MGU's properties theoretically and empirically.", "Two potential bottlenecks on the expressiveness of recurrent neural networks (RNNs) are their ability to store information about the task in their parameters, and to store information about the input history in their units. We show experimentally that all common RNN architectures achieve nearly the same per-task and per-unit capacity bounds with careful training, for a variety of tasks and stacking depths. They can store an amount of task information which is linear in the number of parameters, and is approximately 5 bits per parameter. They can additionally store approximately one real number from their input history per hidden unit. We further find that for several tasks it is the per-task parameter capacity bound that determines performance. These results suggest that many previous results comparing RNN architectures are driven primarily by differences in training effectiveness, rather than differences in capacity. Supporting this observation, we compare training difficulty for several architectures, and show that vanilla RNNs are far more difficult to train, yet have slightly higher capacity. Finally, we propose two novel RNN architectures, one of which is easier to train than the LSTM or GRU for deeply stacked architectures." ] }
1906.08856	2952632952	Learning long-term dependencies still remains difficult for recurrent neural networks (RNNs) despite their success in sequence modeling recently. In this paper, we propose a novel gated RNN structure, which contains only one gate. Hidden states in the proposed grouped distributor unit (GDU) are partitioned into groups. For each group, the proportion of memory to be overwritten in each state transition is limited to a constant and is adaptively distributed to each group member. In other word, every separate group has a fixed overall update rate, yet all units are allowed to have different paces. Information is therefore forced to be latched in a flexible way, which helps the model to capture long-term dependencies in data. Besides having a simpler structure, GDU is demonstrated experimentally to outperform LSTM and GRU on tasks including both pathological problems and natural data set.	El Hihi and Bengio first showed that RNNs can learn both long- and short-term dependencies more easily and efficiently if state units are partitioned into groups with different timescales @cite_12 . The clockwork RNN (CW-RNN) @cite_32 implemented this by assigning each state unit a fixed temporal granularity, making state transition happens only at its prescribed clock rate. It can also be seen as a member of cGAST family. More specifically, a UGRNN with a special gate operator @math in which each gate vector value @math is explicitly scheduled to saturate at either @math or @math . CW-RNN does not suffer from gradient conflict for it inherently has the ability to latch information. However, the clock rate schedule should be tuned for each task.	{ "cite_N": [ "@cite_32", "@cite_12" ], "mid": [ "2138660131", "2099257174" ], "abstract": [ "Sequence prediction and classification are ubiquitous and challenging problems in machine learning that can require identifying complex dependencies between temporally distant inputs. Recurrent Neural Networks (RNNs) have the ability, in theory, to cope with these temporal dependencies by virtue of the short-term memory implemented by their recurrent (feedback) connections. However, in practice they are difficult to train successfully when long-term memory is required. This paper introduces a simple, yet powerful modification to the simple RNN (SRN) architecture, the Clockwork RNN (CW-RNN), in which the hidden layer is partitioned into separate modules, each processing inputs at its own temporal granularity, making computations only at its prescribed clock rate. Rather than making the standard RNN models more complex, CW-RNN reduces the number of SRN parameters, improves the performance significantly in the tasks tested, and speeds up the network evaluation. The network is demonstrated in preliminary experiments involving three tasks: audio signal generation, TIMIT spoken word classification, where it outperforms both SRN and LSTM networks, and online handwriting recognition, where it outperforms SRNs.", "We have already shown that extracting long-term dependencies from sequential data is difficult, both for determimstic dynamical systems such as recurrent networks, and probabilistic models such as hidden Markov models (HMMs) or input output hidden Markov models (IOHMMs). In practice, to avoid this problem, researchers have used domain specific a-priori knowledge to give meaning to the hidden or state variables representing past context. In this paper, we propose to use a more general type of a-priori knowledge, namely that the temporal dependencies are structured hierarchically. This implies that long-term dependencies are represented by variables with a long time scale. This principle is applied to a recurrent network which includes delays and multiple time scales. Experiments confirm the advantages of such structures. A similar approach is proposed for HMMs and IOHMMs." ] }
1906.08905	2951325939	Exploiting different representations, or views, of the same object for better clustering has become very popular these days, which is conventionally called multi-view clustering. Generally, it is essential to measure the importance of each individual view, due to some noises, or inherent capacities in description. Many previous works model the view importance as weight, which is simple but effective empirically. In this paper, instead of following the traditional thoughts, we propose a new weight learning paradigm in context of multi-view clustering in virtue of the idea of re-weighted approach, and we theoretically analyze its working mechanism. Meanwhile, as a carefully achieved example, all of the views are connected by exploring a unified Laplacian rank constrained graph, which will be a representative method to compare with other weight learning approaches in experiments. Furthermore, the proposed weight learning strategy is much suitable for multi-view data, and it can be naturally integrated with many existing clustering learners. According to the numerical experiments, the proposed intrinsic weight learning approach is proved effective and practical to use in multi-view clustering.	Many previous multi-view clustering methods have made attempts to address the view variance problem. In @cite_11 , the authors points out that learning the view importance is very necessary, but they simply resort to the extra prior knowledge in the paper. Some others, such as recent approaches @cite_14 @cite_51 , tackle this problem by modeling the separated noise in each view and learn a shared clean data structure. However, this way is not able to cover the second case of view variance because of the gap of real noises and clustering errors by a weak view. As a matter of fact, measuring the view importance with the weight is direct and useful, and many works prefer this way. For convenience, @cite_19 proposes to roughly compute the weights according to the proportion of the graph volume in each view. Obviously, this strategy is manually intervening and much shallow. Practically, most works would prefer to learn the weights automatically or adaptively. That means they usually optimize the objective and weights simultaneously. In the followings, we particularly introduce several types of how the previous works learn the weights which is closely related our work.	{ "cite_N": [ "@cite_19", "@cite_14", "@cite_51", "@cite_11" ], "mid": [ "2014143538", "201974436", "", "2101324110" ], "abstract": [ "Clustering on multiple views is witnessing increasing interests in both real-world application and machine learning community. A typical application is to discover communities of joint interests in social network, such as Facebook and Twitter. The network can be simply modeled as a graph in which the nodes are the people while the links show relationship between the people. There may exist many relationships between a pair of nodes, such as classmates, collaborators, playmates and so on. It is important to consider how to use these graphs together rather than a single graph if we want to understand the network and their participants effectively. Motivated by the fact, we present a clustering algorithm using spectral analysis in which multiple graphs are considered to get the clusters. Our study can also be considered as an instance of multi-views learning. The experimental results on UCI data set and Corel image data demonstrate the promising results that validate our proposed algorithm.", "Multi-view clustering, which seeks a partition of the data in multiple views that often provide complementary information to each other, has received considerable attention in recent years. In real life clustering problems, the data in each view may have considerable noise. However, existing clustering methods blindly combine the information from multi-view data with possibly considerable noise, which often degrades their performance. In this paper, we propose a novel Markov chain method for Robust Multi-view Spectral Clustering (RMSC). Our method has a flavor of lowrank and sparse decomposition, where we firstly construct a transition probability matrix from each single view, and then use these matrices to recover a shared low-rank transition probability matrix as a crucial input to the standard Markov chain method for clustering. The optimization problem of RMSC has a low-rank constraint on the transition probability matrix, and simultaneously a probabilistic simplex constraint on each of its rows. To solve this challenging optimization problem, we propose an optimization procedure based on the Augmented Lagrangian Multiplier scheme. Experimental results on various real world datasets show that the proposed method has superior performance over several state-of-the-art methods for multi-view clustering.", "", "We propose a spectral clustering algorithm for the multi-view setting where we have access to multiple views of the data, each of which can be independently used for clustering. Our spectral clustering algorithm has a flavor of co-training, which is already a widely used idea in semi-supervised learning. We work on the assumption that the true underlying clustering would assign a point to the same cluster irrespective of the view. Hence, we constrain our approach to only search for the clusterings that agree across the views. Our algorithm does not have any hyperparameters to set, which is a major advantage in unsupervised learning. We empirically compare with a number of baseline methods on synthetic and real-world datasets to show the efficacy of the proposed algorithm." ] }
1906.08905	2951325939	Exploiting different representations, or views, of the same object for better clustering has become very popular these days, which is conventionally called multi-view clustering. Generally, it is essential to measure the importance of each individual view, due to some noises, or inherent capacities in description. Many previous works model the view importance as weight, which is simple but effective empirically. In this paper, instead of following the traditional thoughts, we propose a new weight learning paradigm in context of multi-view clustering in virtue of the idea of re-weighted approach, and we theoretically analyze its working mechanism. Meanwhile, as a carefully achieved example, all of the views are connected by exploring a unified Laplacian rank constrained graph, which will be a representative method to compare with other weight learning approaches in experiments. Furthermore, the proposed weight learning strategy is much suitable for multi-view data, and it can be naturally integrated with many existing clustering learners. According to the numerical experiments, the proposed intrinsic weight learning approach is proved effective and practical to use in multi-view clustering.	To make the weight distribution flater, some works @cite_37 @cite_9 @cite_34 @cite_5 add a norm regularization term, and thus the objective comes to where @math is a non-negative parameter which controls the degree of flatness. When @math , Eq. reduces to Eq. and the best view will be selected. On the contrary, when @math , the equal weights will be obtained. Particularly, when @math is fixed, the derived subproblem is where ( = [ _1 ( x ), _2 ( x ),..., _M ( x ) ] ). This problem can be effectively solved by the algorithm in @cite_38 , and the obtained weights are usually sparse (see the discussion therein.). .	{ "cite_N": [ "@cite_38", "@cite_37", "@cite_9", "@cite_5", "@cite_34" ], "mid": [ "1978259121", "2056649800", "", "2007477772", "1517669128" ], "abstract": [ "We describe efficient algorithms for projecting a vector onto the l1-ball. We present two methods for projection. The first performs exact projection in O(n) expected time, where n is the dimension of the space. The second works on vectors k of whose elements are perturbed outside the l1-ball, projecting in O(k log(n)) time. This setting is especially useful for online learning in sparse feature spaces such as text categorization applications. We demonstrate the merits and effectiveness of our algorithms in numerous batch and online learning tasks. We show that variants of stochastic gradient projection methods augmented with our efficient projection procedures outperform interior point methods, which are considered state-of-the-art optimization techniques. We also show that in online settings gradient updates with l1 projections outperform the exponentiated gradient algorithm while obtaining models with high degrees of sparsity.", "Background Modeling high-dimensional data involving thousands of variables is particularly important for gene expression profiling experiments, nevertheless,it remains a challenging task. One of the challenges is to implement an effective method for selecting a small set of relevant genes, buried in high-dimensional irrelevant noises. RELIEF is a popular and widely used approach for feature selection owing to its low computational cost and high accuracy. However, RELIEF based methods suffer from instability, especially in the presence of noisy and or high-dimensional outliers.", "", "Graph-based approaches have been most successful in semisupervised learning. In this paper, we focus on label propagation in graph-based semisupervised learning. One essential point of label propagation is that the performance is heavily affected by incorporating underlying manifold of given data into the input graph. The other more important point is that in many recent real-world applications, the same instances are represented by multiple heterogeneous data sources. A key challenge under this setting is to integrate different data representations automatically to achieve better predictive performance. In this paper, we address the issue of obtaining the optimal linear combination of multiple different graphs under the label propagation setting. For this problem, we propose a new formulation with the sparsity (in coefficients of graph combination) property which cannot be rightly achieved by any other existing methods. This unique feature provides two important advantages: 1) the improvement of prediction performance by eliminating irrelevant or noisy graphs and 2) the interpretability of results, i.e., easily identifying informative graphs on classification. We propose efficient optimization algorithms for the proposed approach, by which clear interpretations of the mechanism for sparsity is provided. Through various synthetic and two real-world data sets, we empirically demonstrate the advantages of our proposed approach not only in prediction performance but also in graph selection ability.", "In this paper we present an unsupervised method to learn the weights with which the scores of multiple classifiers must be combined in classifier fusion settings. We also introduce a novel metric for ranking instances based on an index which depends upon the rank of weighted scores of test points among the weighted scores of training points. We show that the optimized index can be used for computing measures such as average precision. Unlike most classifier fusion methods where a single weight is learned to weigh all examples our method learns instance-specific weights. The problem is formulated as learning the weight which maximizes a clarity index; subsequently the index itself and the learned weights both are used separately to rank all the test points. Our method gives an unsupervised method of optimizing performance on actual test data, unlike the well known stacking-based methods where optimization is done over a labeled training set. Moreover, we show that our method is tolerant to noisy classifiers and can be used for selecting N-best classifiers." ] }
1906.08905	2951325939	Exploiting different representations, or views, of the same object for better clustering has become very popular these days, which is conventionally called multi-view clustering. Generally, it is essential to measure the importance of each individual view, due to some noises, or inherent capacities in description. Many previous works model the view importance as weight, which is simple but effective empirically. In this paper, instead of following the traditional thoughts, we propose a new weight learning paradigm in context of multi-view clustering in virtue of the idea of re-weighted approach, and we theoretically analyze its working mechanism. Meanwhile, as a carefully achieved example, all of the views are connected by exploring a unified Laplacian rank constrained graph, which will be a representative method to compare with other weight learning approaches in experiments. Furthermore, the proposed weight learning strategy is much suitable for multi-view data, and it can be naturally integrated with many existing clustering learners. According to the numerical experiments, the proposed intrinsic weight learning approach is proved effective and practical to use in multi-view clustering.	An alternative to NR is to utilize the maximum entropy @cite_25 to penalize the weights. It can be described as where @math has the identical effect with @math in Eq. . Many previous works learn the weights in this way, such as @cite_28 @cite_2 . Similarly, when @math is fixed, we give the analytical solution to the corresponding subproblem as which is also known as Gibbs distribution as in @cite_28 . According to Eq. , it can be observed that when ( _v ( x ) _i ( x ) _2 . - _2 ) is very large, @math will be very small. Thus, loosely speaking, this strategy also learns the sparse weights. .	{ "cite_N": [ "@cite_28", "@cite_25", "@cite_2" ], "mid": [ "2106372467", "2032558547", "2397647644" ], "abstract": [ "Fusing multiple information sources can yield significant benefits to successfully accomplish learning tasks. Many studies have focussed on fusing information in supervised learning contexts. We present an approach to utilize multiple information sources in the form of similarity data for unsupervised learning. Based on similarity information, the clustering task is phrased as a non-negative matrix factorization problem of a mixture of similarity measurements. The tradeoff between the informativeness of data sources and the sparseness of their mixture is controlled by an entropy-based weighting mechanism. For the purpose of model selection, a stability-based approach is employed to ensure the selection of the most self-consistent hypothesis. The experiments demonstrate the performance of the method on toy as well as real world data sets.", "", "Multi-view clustering has become a popular clustering technique in recent years due to its ability to analyze data collected from multiple sources or represented by multiple views. In this paper, we propose a novel multi-view clustering approach termed weighted multi-view online competitive clustering (WMLCC). We simultaneously exploit the variable weighting strategy and the online competitive learning in our approach and cast the multi-view clustering problem into an optimization problem. The multi-view clustering result can be obtained by optimizing a new objective function. We conduct Multi-view clustering has become a popular clustering technique in recent years due to its ability to analyze data collected from multiple sources or represented by multiple views. In this paper, we propose a novel multi-view clustering approach termed weighted multi-view online competitive clustering (WMLCC). We simultaneously exploit the variable weighting strategy and the online competitive learning in our approach and cast the multi-view clustering problem into an optimization problem. The multi-view clustering result can be obtained by optimizing a new objective function. We conduct experiments on two real-world multi-view datasets. Experimental results demonstrate the effectiveness and efficiency of our approach." ] }
1906.08905	2951325939	Exploiting different representations, or views, of the same object for better clustering has become very popular these days, which is conventionally called multi-view clustering. Generally, it is essential to measure the importance of each individual view, due to some noises, or inherent capacities in description. Many previous works model the view importance as weight, which is simple but effective empirically. In this paper, instead of following the traditional thoughts, we propose a new weight learning paradigm in context of multi-view clustering in virtue of the idea of re-weighted approach, and we theoretically analyze its working mechanism. Meanwhile, as a carefully achieved example, all of the views are connected by exploring a unified Laplacian rank constrained graph, which will be a representative method to compare with other weight learning approaches in experiments. Furthermore, the proposed weight learning strategy is much suitable for multi-view data, and it can be naturally integrated with many existing clustering learners. According to the numerical experiments, the proposed intrinsic weight learning approach is proved effective and practical to use in multi-view clustering.	Another approach to smoothen the weight distribution is to introduce a parameter as the exponent of each @math where ( _3 > 1 ). Due to the free of the regularization term, numerous multi-view clustering works @cite_8 @cite_42 @cite_9 @cite_2 @cite_31 have adopted it. Fixing @math , the solution of the subproblem can be given as	{ "cite_N": [ "@cite_8", "@cite_9", "@cite_42", "@cite_2", "@cite_31" ], "mid": [ "1564937518", "", "2169529055", "2397647644", "2160914206" ], "abstract": [ "Different features describe different views of visual appearance, multi-view based methods can integrate the information contained in each view and improve the image clustering performance. Most of the existing methods assume that the importance of one type of feature is the same to all the data. However, the visual appearance of images are different, so the description abilities of different features vary with different images. To solve this problem, we propose a group-aware multi-view fusion approach. Images are partitioned into groups which consist of several images sharing similar visual appearance. We assign different weights to evaluate the pairwise similarity between different groups. Then the clustering results and the fusion weights are learned by an iterative optimization procedure. Experimental results indicate that our approach achieves promising clustering performance compared with the existing methods.", "", "Exploiting multiple representations, or views, for the same set of instances within a clustering framework is a popular practice for boosting clustering accuracy. However, some of the available sources may be misleading (due to noise, errors in measurement etc.) in revealing the true structure of the data, thus, their inclusion in the clustering process may have negative influence. This aspect seems to be overlooked in the multi-view literature where all representations are equally considered. In this work, views are expressed in terms of given kernel matrices and a weighted combination of the kernels is learned in parallel to the partitioning. Weights assigned to kernels are indicative of the quality of the corresponding views' information. Additionally, the combination scheme incorporates a parameter that controls the admissible sparsity of the weights to avoid extremes and tailor them to the data. Two efficient iterative algorithms are proposed that alternate between updating the view weights and recomputing the clusters to optimize the intra-cluster variance from different perspectives. The conducted experiments reveal the effectiveness of our methodology compared to other multi-view methods.", "Multi-view clustering has become a popular clustering technique in recent years due to its ability to analyze data collected from multiple sources or represented by multiple views. In this paper, we propose a novel multi-view clustering approach termed weighted multi-view online competitive clustering (WMLCC). We simultaneously exploit the variable weighting strategy and the online competitive learning in our approach and cast the multi-view clustering problem into an optimization problem. The multi-view clustering result can be obtained by optimizing a new objective function. We conduct Multi-view clustering has become a popular clustering technique in recent years due to its ability to analyze data collected from multiple sources or represented by multiple views. In this paper, we propose a novel multi-view clustering approach termed weighted multi-view online competitive clustering (WMLCC). We simultaneously exploit the variable weighting strategy and the online competitive learning in our approach and cast the multi-view clustering problem into an optimization problem. The multi-view clustering result can be obtained by optimizing a new objective function. We conduct experiments on two real-world multi-view datasets. Experimental results demonstrate the effectiveness and efficiency of our approach.", "Multiview clustering partitions a dataset into groups by simultaneously considering multiple representations (views) for the same instances. Hence, the information available in all views is exploited and this may substantially improve the clustering result obtained by using a single representation. Usually, in multiview algorithms all views are considered equally important, something that may lead to bad cluster assignments if a view is of poor quality. To deal with this problem, we propose a method that is built upon exemplar-based mixture models, called convex mixture models (CMMs). More specifically, we present a multiview clustering algorithm, based on training a weighted multiview CMM, that associates a weight with each view and learns these weights automatically. Our approach is computationally efficient and easy to implement, involving simple iterative computations. Experiments with several datasets confirm the advantages of assigning weights to the views and the superiority of our framework over single-view and unweighted multiview CMMs, as well as over another multiview algorithm which is based on kernel canonical correlation analysis." ] }
1811.07746	2901632291	We use multiple measures of graph complexity to evaluate the realism of synthetically-generated networks of human activity, in comparison with several stylized network models as well as a collection of empirical networks from the literature. The synthetic networks are generated by integrating data about human populations from several sources, including the Census, transportation surveys, and geographical data. The resulting networks represent an approximation of daily or weekly human interaction. Our results indicate that the synthetically generated graphs according to our methodology are closer to the real world graphs, as measured across multiple structural measures, than a range of stylized graphs generated using common network models from the literature.	@cite_45 showed that there is a significant difference in how hard it is to shatter social networks as compared to infrastructural networks (transportation, power, and wireless radio networks), by examining the size of the largest component in the network as nodes are deleted in order of (residual) degree. Social networks were shown to be much more robust than the infrastructural networks.	{ "cite_N": [ "@cite_45" ], "mid": [ "2098109664" ], "abstract": [ "Over the last fifteen years, researchers in the Basic and Applied Simulation Science group at Los Alamos National Laboratory have worked with various collaborators on the modeling, simulation, and development of associated decision support tools for understanding large socio-technical systems. The extremely detailed multiscale computer simulations allow individual agents (e.g., people, cars, digital devices) to interact among themselves, as well as with the environment and the networked infrastructure. Such simulations are helpful to policy makers and infrastructure planners who need to answer specific questions. Additionally, our formal results show that simulation-based methods are both necessary and sufficient for understanding the dynamics of such complex systems [4–5]. The mathematical and computational theory views simulations as certain kinds of discrete dynamical systems and provides formal methods for the design, specification, and analysis of such simulations [3–5,11]. Unlike physical systems, socio-technical systems are affected not only by physical laws but also by human behavior, regulatory agencies, and government and private enterprise. The simulation of such systems thus presents novel challenges to researchers. Urban transportation systems constitute a canonical example of the types and levels of interactions that characterize these systems: Traffic rules in distant parts of a city can have an important bearing on traffic congestion downtown, and seemingly “reasonable”" ] }
1811.07746	2901632291	We use multiple measures of graph complexity to evaluate the realism of synthetically-generated networks of human activity, in comparison with several stylized network models as well as a collection of empirical networks from the literature. The synthetic networks are generated by integrating data about human populations from several sources, including the Census, transportation surveys, and geographical data. The resulting networks represent an approximation of daily or weekly human interaction. Our results indicate that the synthetically generated graphs according to our methodology are closer to the real world graphs, as measured across multiple structural measures, than a range of stylized graphs generated using common network models from the literature.	studied differences in social contact networks generated according to different methods @cite_37 and the differences between the networks generated for two different cities (Delhi and Los Angeles) @cite_27 . They used both static and dynamic measures including degree distribution, temporal degree, vulnerability distribution, simulated epidemic curves, and the efficacy of two different epidemic mitigation interventions. Vulnerability is the probability that a node gets infected during an infectious disease outbreak. They showed significant differences due to the network generation methodology as well as across the two cities.	{ "cite_N": [ "@cite_27", "@cite_37" ], "mid": [ "2186454446", "2060161079" ], "abstract": [ "We describe the synthesis of detailed social contact networks of Delhi, India, and Los Angeles, USA, for urban-scale epidemiological simulations. The network synthesis is done by combining information from multiple data sources, since social contact information cannot be obtained through direct surveys. We compare the two networks on various structural and dynamical metrics. Through the comparison between the two cities, we show important similarities and dierences between urban regions in dierent parts of the world.", "Social contact networks play an increasingly important role in computational epidemiology. We focus on massive social contact networks that cover an urban region comprising of millions of individuals and billions of time varying interactions. It is impossible to obtain such networks by simply measuring social interactions. As a result, such networks are often synthesized using a diverse set of real-world data. The synthesis method can be viewed as a complex stochastic process that outputs one realization of such a network. The resulting networks are extremely large, dynamic and unstructured. Any meaningful description of such networks is usually done in terms of structural properties. Building on our earlier work, we synthesize a detailed social contact network for the National Capital Territory (NCT) of India. We first synthesize a social visitation network, representing people visiting locations during different time intervals. We then project it to synthesize a people-people contact network. We are not aware of other works on synthesizing the NCT network. Two important questions arise when synthesizing such massive dynamic social contact networks: (i) how does one compare the networks that span the same region and (ii) when is the synthesized network adequate. To address them, we compute a number of network measurements: some of which are classical, while others capture the semantics of social contact networks. These metrics are used to study the similarities and differences between two networks representing the same urban region. For question (ii), we study our ability to understand the dynamics and control of epidemics. Dynamical measures that capture the joint interaction between the local dynamical process and the network structure are presented and used to analyze the NCT networks." ] }
1811.07789	2901801467	Researchers have observed that Visual Question Answering (VQA) models tend to answer questions by learning statistical biases in the data. For example, their answer to the question "What is the color of the grass?" is usually "Green", whereas a question like "What is the title of the book?" cannot be answered by inferring statistical biases. It is of interest to the community to explicitly discover such biases, both for understanding the behavior of such models, and towards debugging them. Our work address this problem. In a database, we store the words of the question, answer and visual words corresponding to regions of interest in attention maps. By running simple rule mining algorithms on this database, we discover human-interpretable rules which give us unique insight into the behavior of such models. Our results also show examples of unusual behaviors learned by models in attempting VQA tasks.	: The seminal work by @cite_12 suggests that the Machine Learning community does not have a good understanding of what it means to interpret a model. In particular, this work expounds - interpretation of a model's behavior based on some criteria, such as visualizations of gradients @cite_8 or attention maps @cite_16 , the model has been trained. Locally Interpretable Model Agnostic Explanations (LIME), @cite_34 explain a classifier's behavior at a particular point by perturbing the sample and building a linear model using the perturbations and their predictions. A follow up work @cite_24 constructs , which are features such that, in an instance where these features hold, a model's prediction does not change. This work is the most similar prior work to ours, and the authors provide a few results on as well. However, they only assume the existence of a model, and perturb instances of the data, whereas ours assumes the existence of responses to a dataset, but not the model itself. We use standard rule finding algorithms and provide much more detailed results on the problem.	{ "cite_N": [ "@cite_8", "@cite_16", "@cite_24", "@cite_34", "@cite_12" ], "mid": [ "2616247523", "2950178297", "", "2282821441", "2439568532" ], "abstract": [ "We propose a technique for producing \"visual explanations\" for decisions from a large class of CNN-based models, making them more transparent. Our approach - Gradient-weighted Class Activation Mapping (Grad-CAM), uses the gradients of any target concept, flowing into the final convolutional layer to produce a coarse localization map highlighting the important regions in the image for predicting the concept. Unlike previous approaches, GradCAM is applicable to a wide variety of CNN model-families: (1) CNNs with fully-connected layers (e.g. VGG), (2) CNNs used for structured outputs (e.g. captioning), (3) CNNs used in tasks with multimodal inputs (e.g. VQA) or reinforcement learning, without any architectural changes or re-training. We combine GradCAM with fine-grained visualizations to create a high-resolution class-discriminative visualization and apply it to off-the-shelf image classification, captioning, and visual question answering (VQA) models, including ResNet-based architectures. In the context of image classification models, our visualizations (a) lend insights into their failure modes (showing that seemingly unreasonable predictions have reasonable explanations), (b) are robust to adversarial images, (c) outperform previous methods on weakly-supervised localization, (d) are more faithful to the underlying model and (e) help achieve generalization by identifying dataset bias. For captioning and VQA, our visualizations show that even non-attention based models can localize inputs. Finally, we conduct human studies to measure if GradCAM explanations help users establish trust in predictions from deep networks and show that GradCAM helps untrained users successfully discern a \"stronger\" deep network from a \"weaker\" one. Our code is available at this https URL A demo and a video of the demo can be found at this http URL and youtu.be COjUB9Izk6E.", "Inspired by recent work in machine translation and object detection, we introduce an attention based model that automatically learns to describe the content of images. We describe how we can train this model in a deterministic manner using standard backpropagation techniques and stochastically by maximizing a variational lower bound. We also show through visualization how the model is able to automatically learn to fix its gaze on salient objects while generating the corresponding words in the output sequence. We validate the use of attention with state-of-the-art performance on three benchmark datasets: Flickr8k, Flickr30k and MS COCO.", "", "Despite widespread adoption, machine learning models remain mostly black boxes. Understanding the reasons behind predictions is, however, quite important in assessing trust, which is fundamental if one plans to take action based on a prediction, or when choosing whether to deploy a new model. Such understanding also provides insights into the model, which can be used to transform an untrustworthy model or prediction into a trustworthy one. In this work, we propose LIME, a novel explanation technique that explains the predictions of any classifier in an interpretable and faithful manner, by learning an interpretable model locally varound the prediction. We also propose a method to explain models by presenting representative individual predictions and their explanations in a non-redundant way, framing the task as a submodular optimization problem. We demonstrate the flexibility of these methods by explaining different models for text (e.g. random forests) and image classification (e.g. neural networks). We show the utility of explanations via novel experiments, both simulated and with human subjects, on various scenarios that require trust: deciding if one should trust a prediction, choosing between models, improving an untrustworthy classifier, and identifying why a classifier should not be trusted.", "Supervised machine learning models boast remarkable predictive capabilities. But can you trust your model? Will it work in deployment? What else can it tell you about the world? We want models to be not only good, but interpretable. And yet the task of interpretation appears underspecified. Papers provide diverse and sometimes non-overlapping motivations for interpretability, and offer myriad notions of what attributes render models interpretable. Despite this ambiguity, many papers proclaim interpretability axiomatically, absent further explanation. In this paper, we seek to refine the discourse on interpretability. First, we examine the motivations underlying interest in interpretability, finding them to be diverse and occasionally discordant. Then, we address model properties and techniques thought to confer interpretability, identifying transparency to humans and post-hoc explanations as competing notions. Throughout, we discuss the feasibility and desirability of different notions, and question the oft-made assertions that linear models are interpretable and that deep neural networks are not." ] }
1811.07789	2901801467	Researchers have observed that Visual Question Answering (VQA) models tend to answer questions by learning statistical biases in the data. For example, their answer to the question "What is the color of the grass?" is usually "Green", whereas a question like "What is the title of the book?" cannot be answered by inferring statistical biases. It is of interest to the community to explicitly discover such biases, both for understanding the behavior of such models, and towards debugging them. Our work address this problem. In a database, we store the words of the question, answer and visual words corresponding to regions of interest in attention maps. By running simple rule mining algorithms on this database, we discover human-interpretable rules which give us unique insight into the behavior of such models. Our results also show examples of unusual behaviors learned by models in attempting VQA tasks.	: @cite_31 study the behavior of models on the 1.0 dataset. Through a series of experiments, they show that models fail on novel instances, tend to answer after only partially reading the question and fail to change their answers across different images. In @cite_32 , recognizing that deep models seem to use a combination of identifying visual concepts and prediction of answers using biases learned from the data, the authors develop a mechanism to disentangle the two. However, they do not explicitly find a way to discover such biases in the first place. In @cite_21 , the authors introduce a second, more balanced version of the dataset that mitigates biases (especially language based ones) in the original dataset. The resulting balanced dataset is christened 2.0, and is the dataset that our results are reported on. In @cite_18 , the authors balance yes no questions (those which indicate the presence or absence of objects), and propose two new evaluation metrics that compensate for forms of dataset bias.	{ "cite_N": [ "@cite_31", "@cite_21", "@cite_18", "@cite_32" ], "mid": [ "2463267937", "2952228917", "2597425697", "2771951981" ], "abstract": [ "Recently, a number of deep-learning based models have been proposed for the task of Visual Question Answering (VQA). The performance of most models is clustered around 60-70 . In this paper we propose systematic methods to analyze the behavior of these models as a first step towards recognizing their strengths and weaknesses, and identifying the most fruitful directions for progress. We analyze two models, one each from two major classes of VQA models -- with-attention and without-attention and show the similarities and differences in the behavior of these models. We also analyze the winning entry of the VQA Challenge 2016. Our behavior analysis reveals that despite recent progress, today's VQA models are \"myopic\" (tend to fail on sufficiently novel instances), often \"jump to conclusions\" (converge on a predicted answer after 'listening' to just half the question), and are \"stubborn\" (do not change their answers across images).", "Problems at the intersection of vision and language are of significant importance both as challenging research questions and for the rich set of applications they enable. However, inherent structure in our world and bias in our language tend to be a simpler signal for learning than visual modalities, resulting in models that ignore visual information, leading to an inflated sense of their capability. We propose to counter these language priors for the task of Visual Question Answering (VQA) and make vision (the V in VQA) matter! Specifically, we balance the popular VQA dataset by collecting complementary images such that every question in our balanced dataset is associated with not just a single image, but rather a pair of similar images that result in two different answers to the question. Our dataset is by construction more balanced than the original VQA dataset and has approximately twice the number of image-question pairs. Our complete balanced dataset is publicly available at this http URL as part of the 2nd iteration of the Visual Question Answering Dataset and Challenge (VQA v2.0). We further benchmark a number of state-of-art VQA models on our balanced dataset. All models perform significantly worse on our balanced dataset, suggesting that these models have indeed learned to exploit language priors. This finding provides the first concrete empirical evidence for what seems to be a qualitative sense among practitioners. Finally, our data collection protocol for identifying complementary images enables us to develop a novel interpretable model, which in addition to providing an answer to the given (image, question) pair, also provides a counter-example based explanation. Specifically, it identifies an image that is similar to the original image, but it believes has a different answer to the same question. This can help in building trust for machines among their users.", "In visual question answering (VQA), an algorithm must answer text-based questions about images. While multiple datasets for VQA have been created since late 2014, they all have flaws in both their content and the way algorithms are evaluated on them. As a result, evaluation scores are inflated and predominantly determined by answering easier questions, making it difficult to compare different methods. In this paper, we analyze existing VQA algorithms using a new dataset called the Task Driven Image Understanding Challenge (TDIUC), which has over 1.6 million questions organized into 12 different categories. We also introduce questions that are meaningless for a given image to force a VQA system to reason about image content. We propose new evaluation schemes that compensate for over-represented question-types and make it easier to study the strengths and weaknesses of algorithms. We analyze the performance of both baseline and state-of-the-art VQA models, including multi-modal compact bilinear pooling (MCB), neural module networks, and recurrent answering units. Our experiments establish how attention helps certain categories more than others, determine which models work better than others, and explain how simple models (e.g. MLP) can surpass more complex models (MCB) by simply learning to answer large, easy question categories.", "A number of studies have found that today's Visual Question Answering (VQA) models are heavily driven by superficial correlations in the training data and lack sufficient image grounding. To encourage development of models geared towards the latter, we propose a new setting for VQA where for every question type, train and test sets have different prior distributions of answers. Specifically, we present new splits of the VQA v1 and VQA v2 datasets, which we call Visual Question Answering under Changing Priors (VQA-CP v1 and VQA-CP v2 respectively). First, we evaluate several existing VQA models under this new setting and show that their performance degrades significantly compared to the original VQA setting. Second, we propose a novel Grounded Visual Question Answering model (GVQA) that contains inductive biases and restrictions in the architecture specifically designed to prevent the model from 'cheating' by primarily relying on priors in the training data. Specifically, GVQA explicitly disentangles the recognition of visual concepts present in the image from the identification of plausible answer space for a given question, enabling the model to more robustly generalize across different distributions of answers. GVQA is built off an existing VQA model -- Stacked Attention Networks (SAN). Our experiments demonstrate that GVQA significantly outperforms SAN on both VQA-CP v1 and VQA-CP v2 datasets. Interestingly, it also outperforms more powerful VQA models such as Multimodal Compact Bilinear Pooling (MCB) in several cases. GVQA offers strengths complementary to SAN when trained and evaluated on the original VQA v1 and VQA v2 datasets. Finally, GVQA is more transparent and interpretable than existing VQA models." ] }
1811.07738	2900872212	In this paper, we present a novel neural network architecture for retinal vessel segmentation that improves over the state of the art on two benchmark datasets, is the first to run in real time on high resolution images, and its small memory and processing requirements make it deployable in mobile and embedded systems. The M2U-Net has a new encoder-decoder architecture that is inspired by the U-Net. It adds pretrained components of MobileNetV2 in the encoder part and novel contractive bottleneck blocks in the decoder part that, combined with bilinear upsampling, drastically reduce the parameter count to 0.55M compared to 31.03M in the original U-Net. We have evaluated its performance against a wide body of previously published results on three public datasets. On two of them, the M2U-Net achieves new state-of-the-art performance by a considerable margin. When implemented on a GPU, our method is the first to achieve real-time inference speeds on high-resolution fundus images. We also implemented our proposed network on an ARM-based embedded system where it segments images in between 0.6 and 15 sec, depending on the resolution. Thus, the M2U-Net enables a number of applications of retinal vessel structure extraction, such as early diagnosis of eye diseases, retinal biometric authentication systems, and robot assisted microsurgery.	The task of retinal blood vessel segmentation falls into the computer vision subcategory of semantic segmentation, which in recent years has seen tremendous improvements in performance thanks to the introduction of novel deep neural network architectures @cite_30 @cite_25 @cite_32 . Similarly, recent state-of-the-art methods in retinal blood vessel segmentation that focus on segmentation quality are dominated by various variations of deep neural networks @cite_12 @cite_35 @cite_10 .	{ "cite_N": [ "@cite_30", "@cite_35", "@cite_32", "@cite_10", "@cite_25", "@cite_12" ], "mid": [ "2952632681", "2513326255", "2952232639", "2802388893", "", "2327793514" ], "abstract": [ "Convolutional networks are powerful visual models that yield hierarchies of features. We show that convolutional networks by themselves, trained end-to-end, pixels-to-pixels, exceed the state-of-the-art in semantic segmentation. Our key insight is to build \"fully convolutional\" networks that take input of arbitrary size and produce correspondingly-sized output with efficient inference and learning. We define and detail the space of fully convolutional networks, explain their application to spatially dense prediction tasks, and draw connections to prior models. We adapt contemporary classification networks (AlexNet, the VGG net, and GoogLeNet) into fully convolutional networks and transfer their learned representations by fine-tuning to the segmentation task. We then define a novel architecture that combines semantic information from a deep, coarse layer with appearance information from a shallow, fine layer to produce accurate and detailed segmentations. Our fully convolutional network achieves state-of-the-art segmentation of PASCAL VOC (20 relative improvement to 62.2 mean IU on 2012), NYUDv2, and SIFT Flow, while inference takes one third of a second for a typical image.", "This paper presents Deep Retinal Image Understanding (DRIU), a unified framework of retinal image analysis that provides both retinal vessel and optic disc segmentation. We make use of deep Convolutional Neural Networks (CNNs), which have proven revolutionary in other fields of computer vision such as object detection and image classification, and we bring their power to the study of eye fundus images. DRIU uses a base network architecture on which two set of specialized layers are trained to solve both the retinal vessel and optic disc segmentation. We present experimental validation, both qualitative and quantitative, in four public datasets for these tasks. In all of them, DRIU presents super-human performance, that is, it shows results more consistent with a gold standard than a second human annotator used as control.", "There is large consent that successful training of deep networks requires many thousand annotated training samples. In this paper, we present a network and training strategy that relies on the strong use of data augmentation to use the available annotated samples more efficiently. The architecture consists of a contracting path to capture context and a symmetric expanding path that enables precise localization. We show that such a network can be trained end-to-end from very few images and outperforms the prior best method (a sliding-window convolutional network) on the ISBI challenge for segmentation of neuronal structures in electron microscopic stacks. Using the same network trained on transmitted light microscopy images (phase contrast and DIC) we won the ISBI cell tracking challenge 2015 in these categories by a large margin. Moreover, the network is fast. Segmentation of a 512x512 image takes less than a second on a recent GPU. The full implementation (based on Caffe) and the trained networks are available at this http URL .", "Objective: Deep learning based methods for retinal vessel segmentation are usually trained based on pixel-wise losses, which treat all vessel pixels with equal importance in pixel-to-pixel matching between a predicted probability map and the corresponding manually annotated segmentation. However, due to the highly imbalanced pixel ratio between thick and thin vessels in fundus images, a pixel-wise loss would limit deep learning models to learn features for accurate segmentation of thin vessels, which is an important task for clinical diagnosis of eye-related diseases. Methods: In this paper, we propose a new segment-level loss which emphasizes more on the thickness consistency of thin vessels in the training process. By jointly adopting both the segment-level and the pixel-wise losses, the importance between thick and thin vessels in the loss calculation would be more balanced. As a result, more effective features can be learned for vessel segmentation without increasing the overall model complexity. Results: Experimental results on public data sets demonstrate that the model trained by the joint losses outperforms the current state-of-the-art methods in both separate-training and cross-training evaluations. Conclusion: Compared to the pixel-wise loss, utilizing the proposed joint-loss framework is able to learn more distinguishable features for vessel segmentation. In addition, the segment-level loss can bring consistent performance improvement for both deep and shallow network architectures. Significance: The findings from this study of using joint losses can be applied to other deep learning models for performance improvement without significantly changing the network architectures.", "", "The condition of the vascular network of human eye is an important diagnostic factor in ophthalmology. Its segmentation in fundus imaging is a nontrivial task due to variable size of vessels, relatively low contrast, and potential presence of pathologies like microaneurysms and hemorrhages. Many algorithms, both unsupervised and supervised, have been proposed for this purpose in the past. We propose a supervised segmentation technique that uses a deep neural network trained on a large (up to 400 @math 000) sample of examples preprocessed with global contrast normalization, zero-phase whitening, and augmented using geometric transformations and gamma corrections. Several variants of the method are considered, including structured prediction, where a network classifies multiple pixels simultaneously. When applied to standard benchmarks of fundus imaging, the DRIVE, STARE, and CHASE databases, the networks significantly outperform the previous algorithms on the area under ROC curve measure (up to @math ) and accuracy of classification (up to @math ). The method is also resistant to the phenomenon of central vessel reflex, sensitive in detection of fine vessels ( @math ), and fares well on pathological cases." ] }

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.