aid
stringlengths 9
15
| mid
stringlengths 7
10
| abstract
stringlengths 78
2.56k
| related_work
stringlengths 92
1.77k
| ref_abstract
dict |
|---|---|---|---|---|
1906.09248
|
2949366831
|
Machine Learning based Quality of Experience (QoE) models potentially suffer from over-fitting due to limitations including low data volume, and limited participant profiles. This prevents models from becoming generic. Consequently, these trained models may under-perform when tested outside the experimented population. One reason for the limited datasets, which we refer in this paper as small QoE data lakes, is due to the fact that often these datasets potentially contain user sensitive information and are only collected throughout expensive user studies with special user consent. Thus, sharing of datasets amongst researchers is often not allowed. In recent years, privacy preserving machine learning models have become important and so have techniques that enable model training without sharing datasets but instead relying on secure communication protocols. Following this trend, in this paper, we present Round-Robin based Collaborative Machine Learning model training, where the model is trained in a sequential manner amongst the collaborated partner nodes. We benchmark this work using our customized Federated Learning mechanism as well as conventional Centralized and Isolated Learning methods.
|
ML algorithms such as Decision Trees, Random Forests are a few of the most commonly used techniques in the QoE literature , @cite_11 . Support Vector Machines (SVM) have been used earlier in QoE Modeling as they often perform well in small datasets @cite_9 . These models are hard to use for Collaborative Learning as a continuation of training after a model transfer is a challenge. Especially, when larger datasets are used, then there are better alternatives such as Neural Networks, whose weights can be updated using Collaborative Learning techniques.
|
{
"cite_N": [
"@cite_9",
"@cite_11"
],
"mid": [
"2025518782",
"1498320273"
],
"abstract": [
"This paper addresses the challenge of assessing and modeling Quality of Experience (QoE) for online video services that are based on TCP-streaming. We present a dedicated QoE model for You Tube that takes into account the key influence factors (such as stalling events caused by network bottlenecks) that shape quality perception of this service. As second contribution, we propose a generic subjective QoE assessment methodology for multimedia applications (like online video) that is based on crowd sourcing - a highly cost-efficient, fast and flexible way of conducting user experiments. We demonstrate how our approach successfully leverages the inherent strengths of crowd sourcing while addressing critical aspects such as the reliability of the experimental data obtained. Our results suggest that, crowd sourcing is a highly effective QoE assessment method not only for online video, but also for a wide range of other current and future Internet applications.",
"The machine learning provides a theoretical and methodological framework to quantify the relationship between user OoE (Quality of Experience) and network QoS (Quality of Service). This paper presents an overview of QoE-QoS correlation models based on machine learning techniques. According to the learning type, we propose a categorization of correlation models. For each category, we review the main existing works by citing deployed learning methods and model parameters (QoE measurement, QoS parameters and service type). Moreover, the survey will provide researchers with the latest trends and findings in this field."
]
}
|
1811.09971
|
2901359197
|
Recently, graph Convolutional Neural Networks (graph CNNs) have been widely used for graph data representation and semi-supervised learning tasks. However, existing graph CNNs generally use a fixed graph which may be not optimal for semi-supervised learning tasks. In this paper, we propose a novel Graph Learning-Convolutional Network (GLCN) for graph data representation and semi-supervised learning. The aim of GLCN is to learn an optimal graph structure that best serves graph CNNs for semi-supervised learning by integrating both graph learning and graph convolution together in a unified network architecture. The main advantage is that in GLCN, both given labels and the estimated labels are incorporated and thus can provide useful 'weakly' supervised information to refine (or learn) the graph construction and also to facilitate the graph convolution operation in GLCN for unknown label estimation. Experimental results on seven benchmarks demonstrate that GLCN significantly outperforms state-of-the-art traditional fixed structure based graph CNNs.
|
Recently, graph convolutional network (GCN) @cite_13 @cite_8 has been commonly used to address structured graph data. In this section, we briefly review GCN based semi-supervised learning proposed in @cite_8 .
|
{
"cite_N": [
"@cite_13",
"@cite_8"
],
"mid": [
"2963984147",
"2519887557"
],
"abstract": [
"We present diffusion-convolutional neural networks (DCNNs), a new model for graph-structured data. Through the introduction of a diffusion-convolution operation, we show how diffusion-based representations can be learned from graph-structured data and used as an effective basis for node classification. DCNNs have several attractive qualities, including a latent representation for graphical data that is invariant under isomorphism, as well as polynomial-time prediction and learning that can be represented as tensor operations and efficiently implemented on a GPU. Through several experiments with real structured datasets, we demonstrate that DCNNs are able to outperform probabilistic relational models and kernel-on-graph methods at relational node classification tasks.",
"We present a scalable approach for semi-supervised learning on graph-structured data that is based on an efficient variant of convolutional neural networks which operate directly on graphs. We motivate the choice of our convolutional architecture via a localized first-order approximation of spectral graph convolutions. Our model scales linearly in the number of graph edges and learns hidden layer representations that encode both local graph structure and features of nodes. In a number of experiments on citation networks and on a knowledge graph dataset we demonstrate that our approach outperforms related methods by a significant margin."
]
}
|
1811.09971
|
2901359197
|
Recently, graph Convolutional Neural Networks (graph CNNs) have been widely used for graph data representation and semi-supervised learning tasks. However, existing graph CNNs generally use a fixed graph which may be not optimal for semi-supervised learning tasks. In this paper, we propose a novel Graph Learning-Convolutional Network (GLCN) for graph data representation and semi-supervised learning. The aim of GLCN is to learn an optimal graph structure that best serves graph CNNs for semi-supervised learning by integrating both graph learning and graph convolution together in a unified network architecture. The main advantage is that in GLCN, both given labels and the estimated labels are incorporated and thus can provide useful 'weakly' supervised information to refine (or learn) the graph construction and also to facilitate the graph convolution operation in GLCN for unknown label estimation. Experimental results on seven benchmarks demonstrate that GLCN significantly outperforms state-of-the-art traditional fixed structure based graph CNNs.
|
Let @math be the collection of @math data vectors in @math dimension. Let @math be the graph representation of @math with @math encoding the pairwise relationship (such as similarities, neighbors) among data @math . GCN contains one input layer, several propagation (hidden) layers and one final perceptron layer @cite_8 . Given an input @math and graph @math , GCN conducts the following layer-wise propagation in hidden layers as @cite_8 , where @math . @math is a diagonal matrix with @math and @math is a layer-specific weight matrix needing to be trained. @math denotes an activation function, such as @math , and @math denotes the output of activations in the @math -th layer.
|
{
"cite_N": [
"@cite_8"
],
"mid": [
"2519887557"
],
"abstract": [
"We present a scalable approach for semi-supervised learning on graph-structured data that is based on an efficient variant of convolutional neural networks which operate directly on graphs. We motivate the choice of our convolutional architecture via a localized first-order approximation of spectral graph convolutions. Our model scales linearly in the number of graph edges and learns hidden layer representations that encode both local graph structure and features of nodes. In a number of experiments on citation networks and on a knowledge graph dataset we demonstrate that our approach outperforms related methods by a significant margin."
]
}
|
1811.10280
|
2900742222
|
This paper addresses the challenge of humanoid robot teleoperation in a natural indoor environment via a Brain-Computer Interface (BCI). We leverage deep Convolutional Neural Network (CNN) based image and signal understanding to facilitate both real-time object detection and dry-Electroencephalography (EEG) based human cortical brain bio-signals decoding. We employ recent advances in dry-EEG technology to stream and collect the cortical waveforms from subjects while they fixate on variable Steady State Visual Evoked Potential (SSVEP) stimuli generated directly from the environment the robot is navigating. To these ends, we propose the use of novel variable BCI stimuli by utilising the real-time video streamed via the on-board robot camera as visual input for SSVEP, where the CNN detected natural scene objects are altered and flickered with differing frequencies (10Hz, 12Hz and 15Hz). These stimuli are not akin to traditional stimuli - as both the dimensions of the flicker regions and their on-screen position changes depending on the scene objects detected. Onscreen object selection via such a dry-EEG enabled SSVEP methodology, facilitates the on-line decoding of human cortical brain signals, via a specialised secondary CNN, directly into teleoperation robot commands (approach object, move in a specific direction: right, left or back). This SSVEP decoding model is trained via a priori offline experimental data in which very similar visual input is present for all subjects. The resulting classification demonstrates high performance with mean accuracy of 85 for the real-time robot navigation experiment across multiple test subjects.
|
There are two notable studies that have integrated object detection and recognition @cite_29 @cite_1 . In @cite_29 , the authors used seven different frequencies to navigate a mobile robot to a storage rack to grasp an object and delivered into a dustbin with an average mean accuracy of 89.4 The authors in @cite_25 used SSVEP with hybrid-mask feature in which a 3D textured model were rendered and flickered on certain scene objects. In this case, three similar cans which are recognised offline. Subjects for this study teleoperated a humanoid robot HRP-2 (located in Japan from Italy) to control the robot to grasp a can from a table and navigate the robot to a second table where the robot need to drop the can on a marked target.
|
{
"cite_N": [
"@cite_29",
"@cite_1",
"@cite_25"
],
"mid": [
"2738126305",
"",
"2476739747"
],
"abstract": [
"Brain-computer interface (BCI) systems can translate the human mind into control commands, which makes it feasible to improve the life quality of physically challenged people. However, in real-life situations, it is still difficult for users to utilize robots to provide basic services with BCI systems. We aimed to propose a BCI-based system with a visual servo module to operate a service robot. We recorded single-channel steady-state visual evoked potentials (SSVEP) as input signals for the BCI system of this study. The visual stimuli for inducing SSVEP were modulated at seven different frequencies with the sampled sinusoidal method. Correspondingly, this SSVEP-based BCI system can generate seven control commands for the operation of the service robot, which can provide three fundamental services: mobility, manipulation, and delivery. The visual servo module was established to reduce the burden of users and accelerate service procedures. To evaluate the performance of this system, subjects were recruited to participate in the experiments. All the participants succeed in operating the robot to provide the basic services. According to the experimental results, this SSVEP-based BCI system that incorporates the visual servo module can be effectively used to operate service robots with reduced number of channels and increased ability to perform multiple tasks.",
"",
"The efficient control of our body and successful interaction with the environment are possible through the integration of multisensory information. Brain–computer interface (BCI) may allow people with sensorimotor disorders to actively interact in the world. In this study, visual information was paired with auditory feedback to improve the BCI control of a humanoid surrogate. Healthy and spinal cord injured (SCI) people were asked to embody a humanoid robot and complete a pick-and-place task by means of a visual evoked potentials BCI system. Participants observed the remote environment from the robot’s perspective through a head mounted display. Human-footsteps and computer-beep sounds were used as synchronous asynchronous auditory feedback. Healthy participants achieved better placing accuracy when listening to human footstep sounds relative to a computer-generated sound. SCI people demonstrated more difficulty in steering the robot during asynchronous auditory feedback conditions. Importantly, subjective reports highlighted that the BCI mask overlaying the display did not limit the observation of the scenario and the feeling of being in control of the robot. Overall, the data seem to suggest that sensorimotor-related information may improve the control of external devices. Further studies are required to understand how the contribution of residual sensory channels could improve the reliability of BCI systems."
]
}
|
1811.09998
|
2901505625
|
Typically, the deployment of face recognition models in the wild needs to identify low-resolution faces with extremely low computational cost. To address this problem, a feasible solution is compressing a complex face model to achieve higher speed and lower memory at the cost of minimal performance drop. Inspired by that, this paper proposes a learning approach to recognize low-resolution faces via selective knowledge distillation. In this approach, a two-stream convolutional neural network (CNN) is first initialized to recognize high-resolution faces and resolution-degraded faces with a teacher stream and a student stream, respectively. The teacher stream is represented by a complex CNN for high-accuracy recognition, and the student stream is represented by a much simpler CNN for low-complexity recognition. To avoid significant performance drop at the student stream, we then selectively distil the most informative facial features from the teacher stream by solving a sparse graph optimization problem, which are then used to regularize the fine-tuning process of the student stream. In this way, the student stream is actually trained by simultaneously handling two tasks with limited computational resources: approximating the most informative facial cues via feature regression, and recovering the missing facial cues via low-resolution face classification. Experimental results show that the student stream performs impressively in recognizing low-resolution faces and costs only 0.15-MB memory and runs at 418 faces per second on CPU and 9433 faces per second on GPU.
|
Recently, the general face recognition technique has evolved from the classic shallow frameworks @cite_48 @cite_41 to deep ones @cite_44 @cite_7 @cite_9 @cite_18 @cite_0 @cite_34 @cite_1 with impressive performance improvements. For the deep approaches, a key factor to distinguish them is the loss functions they adopted. For example, DeepFace @cite_44 is an early attempt to ensemble Convolutional Neural Networks (CNNs) by building 3D faces with identification loss. After that, various loss functions have been proposed for training face recognition CNNs, such as triplet loss @cite_9 @cite_18 , center loss @cite_7 and range loss @cite_34 . In @cite_1 , the tasks of identifying faces and their attributes were simultaneously considered to enhance the recognition performance. For the DeepID series, several small CNNs using different facial patches were first separately trained in @cite_65 , and its subsequent works incorporate face verification signals @cite_2 and change the base networks @cite_27 to increase accuracy.
|
{
"cite_N": [
"@cite_18",
"@cite_7",
"@cite_41",
"@cite_48",
"@cite_9",
"@cite_1",
"@cite_65",
"@cite_44",
"@cite_0",
"@cite_27",
"@cite_2",
"@cite_34"
],
"mid": [
"2325939864",
"2520774990",
"2121812409",
"1545641654",
"2096733369",
"2738069262",
"1998808035",
"2145287260",
"2584229793",
"2140609507",
"2144172034",
"2557839538"
],
"abstract": [
"The goal of this paper is face recognition – from either a single photograph or from a set of faces tracked in a video. Recent progress in this area has been due to two factors: (i) end to end learning for the task using a convolutional neural network (CNN), and (ii) the availability of very large scale training datasets. We make two contributions: first, we show how a very large scale dataset (2.6M images, over 2.6K people) can be assembled by a combination of automation and human in the loop, and discuss the trade off between data purity and time; second, we traverse through the complexities of deep network training and face recognition to present methods and procedures to achieve comparable state of the art results on the standard LFW and YTF face benchmarks.",
"Convolutional neural networks (CNNs) have been widely used in computer vision community, significantly improving the state-of-the-art. In most of the available CNNs, the softmax loss function is used as the supervision signal to train the deep model. In order to enhance the discriminative power of the deeply learned features, this paper proposes a new supervision signal, called center loss, for face recognition task. Specifically, the center loss simultaneously learns a center for deep features of each class and penalizes the distances between the deep features and their corresponding class centers. More importantly, we prove that the proposed center loss function is trainable and easy to optimize in the CNNs. With the joint supervision of softmax loss and center loss, we can train a robust CNNs to obtain the deep features with the two key learning objectives, inter-class dispension and intra-class compactness as much as possible, which are very essential to face recognition. It is encouraging to see that our CNNs (with such joint supervision) achieve the state-of-the-art accuracy on several important face recognition benchmarks, Labeled Faces in the Wild (LFW), YouTube Faces (YTF), and MegaFace Challenge. Especially, our new approach achieves the best results on MegaFace (the largest public domain face benchmark) under the protocol of small training set (contains under 500000 images and under 20000 persons), significantly improving the previous results and setting new state-of-the-art for both face recognition and face verification tasks.",
"Many current face recognition algorithms perform badly when the lighting or pose of the probe and gallery images differ. In this paper we present a novel algorithm designed for these conditions. We describe face data as resulting from a generative model which incorporates both within-individual and between-individual variation. In recognition we calculate the likelihood that the differences between face images are entirely due to within-individual variability. We extend this to the non-linear case where an arbitrary face manifold can be described and noise is position-dependent. We also develop a \"tied\" version of the algorithm that allows explicit comparison across quite different viewing conditions. We demonstrate that our model produces state of the art results for (i) frontal face recognition (ii) face recognition under varying pose.",
"In this work, we present a novel approach to face recognition which considers both shape and texture information to represent face images. The face area is first divided into small regions from which Local Binary Pattern (LBP) histograms are extracted and concatenated into a single, spatially enhanced feature histogram efficiently representing the face image. The recognition is performed using a nearest neighbour classifier in the computed feature space with Chi square as a dissimilarity measure. Extensive experiments clearly show the superiority of the proposed scheme over all considered methods (PCA, Bayesian Intra extrapersonal Classifier and Elastic Bunch Graph Matching) on FERET tests which include testing the robustness of the method against different facial expressions, lighting and aging of the subjects. In addition to its efficiency, the simplicity of the proposed method allows for very fast feature extraction.",
"Despite significant recent advances in the field of face recognition [10, 14, 15, 17], implementing face verification and recognition efficiently at scale presents serious challenges to current approaches. In this paper we present a system, called FaceNet, that directly learns a mapping from face images to a compact Euclidean space where distances directly correspond to a measure of face similarity. Once this space has been produced, tasks such as face recognition, verification and clustering can be easily implemented using standard techniques with FaceNet embeddings as feature vectors.",
"Deep learning has achieved great success in face recognition, however deep-learned features still have limited invariance to strong intra-personal variations such as large pose changes. It is observed that some facial attributes (e.g. eyebrow thickness, gender) are robust to such variations. We present the first work to systematically explore how the fusion of face recognition features (FRF) and facial attribute features (FAF) can enhance face recognition performance in various challenging scenarios. Despite the promise of FAF, we find that in practice existing fusion methods fail to leverage FAF to boost face recognition performance in some challenging scenarios. Thus, we develop a powerful tensor-based framework which formulates feature fusion as a tensor optimisation problem. It is nontrivial to directly optimise this tensor due to the large number of parameters to optimise. To solve this problem, we establish a theoretical equivalence between low-rank tensor optimisation and a two-stream gated neural network. This equivalence allows tractable learning using standard neural network optimisation tools, leading to accurate and stable optimisation. Experimental results show the fused feature works better than individual features, thus proving for the first time that facial attributes aid face recognition. We achieve state-of-the-art performance on three popular databases: MultiPIE (cross pose, lighting and expression), CASIA NIR-VIS2.0 (cross-modality environment) and LFW (uncontrolled environment).",
"This paper proposes to learn a set of high-level feature representations through deep learning, referred to as Deep hidden IDentity features (DeepID), for face verification. We argue that DeepID can be effectively learned through challenging multi-class face identification tasks, whilst they can be generalized to other tasks (such as verification) and new identities unseen in the training set. Moreover, the generalization capability of DeepID increases as more face classes are to be predicted at training. DeepID features are taken from the last hidden layer neuron activations of deep convolutional networks (ConvNets). When learned as classifiers to recognize about 10, 000 face identities in the training set and configured to keep reducing the neuron numbers along the feature extraction hierarchy, these deep ConvNets gradually form compact identity-related features in the top layers with only a small number of hidden neurons. The proposed features are extracted from various face regions to form complementary and over-complete representations. Any state-of-the-art classifiers can be learned based on these high-level representations for face verification. 97:45 verification accuracy on LFW is achieved with only weakly aligned faces.",
"In modern face recognition, the conventional pipeline consists of four stages: detect => align => represent => classify. We revisit both the alignment step and the representation step by employing explicit 3D face modeling in order to apply a piecewise affine transformation, and derive a face representation from a nine-layer deep neural network. This deep network involves more than 120 million parameters using several locally connected layers without weight sharing, rather than the standard convolutional layers. Thus we trained it on the largest facial dataset to-date, an identity labeled dataset of four million facial images belonging to more than 4, 000 identities. The learned representations coupling the accurate model-based alignment with the large facial database generalize remarkably well to faces in unconstrained environments, even with a simple classifier. Our method reaches an accuracy of 97.35 on the Labeled Faces in the Wild (LFW) dataset, reducing the error of the current state of the art by more than 27 , closely approaching human-level performance.",
"The 3D shapes of faces are well known to be discriminative. Yet despite this, they are rarely used for face recognition and always under controlled viewing conditions. We claim that this is a symptom of a serious but often overlooked problem with existing methods for single view 3D face reconstruction: when applied in the wild, their 3D estimates are either unstable and change for different photos of the same subject or they are over-regularized and generic. In response, we describe a robust method for regressing discriminative 3D morphable face models (3DMM). We use a convolutional neural network (CNN) to regress 3DMM shape and texture parameters directly from an input photo. We overcome the shortage of training data required for this purpose by offering a method for generating huge numbers of labeled examples. The 3D estimates produced by our CNN surpass state of the art accuracy on the MICC data set. Coupled with a 3D-3D face matching pipeline, we show the first competitive face recognition results on the LFW, YTF and IJB-A benchmarks using 3D face shapes as representations, rather than the opaque deep feature vectors used by other modern systems.",
"The state-of-the-art of face recognition has been significantly advanced by the emergence of deep learning. Very deep neural networks recently achieved great success on general object recognition because of their superb learning capacity. This motivates us to investigate their effectiveness on face recognition. This paper proposes two very deep neural network architectures, referred to as DeepID3, for face recognition. These two architectures are rebuilt from stacked convolution and inception layers proposed in VGG net and GoogLeNet to make them suitable to face recognition. Joint face identification-verification supervisory signals are added to both intermediate and final feature extraction layers during training. An ensemble of the proposed two architectures achieves 99.53 LFW face verification accuracy and 96.0 LFW rank-1 face identification accuracy, respectively. A further discussion of LFW face verification result is given in the end.",
"The key challenge of face recognition is to develop effective feature representations for reducing intra-personal variations while enlarging inter-personal differences. In this paper, we show that it can be well solved with deep learning and using both face identification and verification signals as supervision. The Deep IDentification-verification features (DeepID2) are learned with carefully designed deep convolutional networks. The face identification task increases the inter-personal variations by drawing DeepID2 features extracted from different identities apart, while the face verification task reduces the intra-personal variations by pulling DeepID2 features extracted from the same identity together, both of which are essential to face recognition. The learned DeepID2 features can be well generalized to new identities unseen in the training data. On the challenging LFW dataset [11], 99.15 face verification accuracy is achieved. Compared with the best previous deep learning result [20] on LFW, the error rate has been significantly reduced by 67 .",
"Convolutional neural networks have achieved great improvement on face recognition in recent years because of its extraordinary ability in learning discriminative features of people with different identities. To train such a well-designed deep network, tremendous amounts of data is indispensable. Long tail distribution specifically refers to the fact that a small number of generic entities appear frequently while other objects far less existing. Considering the existence of long tail distribution of the real world data, large but uniform distributed data are usually hard to retrieve. Empirical experiences and analysis show that classes with more samples will pose greater impact on the feature learning process and inversely cripple the whole models feature extracting ability on tail part data. Contrary to most of the existing works that alleviate this problem by simply cutting the tailed data for uniform distributions across the classes, this paper proposes a new loss function called range loss to effectively utilize the whole long tailed data in training process. More specifically, range loss is designed to reduce overall intra-personal variations while enlarging inter-personal differences within one mini-batch simultaneously when facing even extremely unbalanced data. The optimization objective of range loss is the @math greatest range's harmonic mean values in one class and the shortest inter-class distance within one batch. Extensive experiments on two famous and challenging face recognition benchmarks (Labeled Faces in the Wild (LFW) and YouTube Faces (YTF) not only demonstrate the effectiveness of the proposed approach in overcoming the long tail effect but also show the good generalization ability of the proposed approach."
]
}
|
1811.09998
|
2901505625
|
Typically, the deployment of face recognition models in the wild needs to identify low-resolution faces with extremely low computational cost. To address this problem, a feasible solution is compressing a complex face model to achieve higher speed and lower memory at the cost of minimal performance drop. Inspired by that, this paper proposes a learning approach to recognize low-resolution faces via selective knowledge distillation. In this approach, a two-stream convolutional neural network (CNN) is first initialized to recognize high-resolution faces and resolution-degraded faces with a teacher stream and a student stream, respectively. The teacher stream is represented by a complex CNN for high-accuracy recognition, and the student stream is represented by a much simpler CNN for low-complexity recognition. To avoid significant performance drop at the student stream, we then selectively distil the most informative facial features from the teacher stream by solving a sparse graph optimization problem, which are then used to regularize the fine-tuning process of the student stream. In this way, the student stream is actually trained by simultaneously handling two tasks with limited computational resources: approximating the most informative facial cues via feature regression, and recovering the missing facial cues via low-resolution face classification. Experimental results show that the student stream performs impressively in recognizing low-resolution faces and costs only 0.15-MB memory and runs at 418 faces per second on CPU and 9433 faces per second on GPU.
|
Typically, there are two ways for low-resolution face recognition. The hallucination category aims to reconstruct high-resolution faces before recognition, while the embedding category proposes extracting features directly from low-resolution faces via the embedding schema. In the hallucination category, Kolouri al @cite_8 constructed a nonlinear Lagrangian model of high-resolution facial appearance and then found the model parameters that best fit the low-resolution faces. Jian al @cite_59 proposed a framework based on singular value decomposition and performed face hallucination and recognition simultaneously. In @cite_55 , a joint face hallucination and recognition framework was proposed based on sparse representation. This framework can synthesize person-specific low-resolution faces for recognition. In @cite_19 , a system was proposed to recognize faces by using sparse representation with the specific dictionary involving many natural and facial images. Moreover, deep models like @cite_47 and @cite_63 can generate extremely realistic high-resolution images from low-resolution faces. However, the speed of such hallucination or super-resolution based approaches may be a little slow due to the complex high-resolution face reconstruction process, which hinders their direct deployment in real-world scenarios with limited computational resources.
|
{
"cite_N": [
"@cite_8",
"@cite_55",
"@cite_19",
"@cite_59",
"@cite_63",
"@cite_47"
],
"mid": [
"1913007689",
"1509693426",
"2440438634",
"2067023690",
"2963470893",
"54257720"
],
"abstract": [
"Extracting high-resolution information from highly degraded facial images is an important problem with several applications in science and technology. Here we describe a single frame super resolution technique that uses a transport-based formulation of the problem. The method consists of a training and a testing phase. In the training phase, a nonlinear Lagrangian model of high-resolution facial appearance is constructed fully automatically. In the testing phase, the resolution of a degraded image is enhanced by finding the model parameters that best fit the given low resolution data. We test the approach on two face datasets, namely the extended Yale Face Database B and the AR face datasets, and compare it to state of the art methods. The proposed method outperforms existing solutions in problems related to enhancing images of very low resolution.",
"In real-world video surveillance applications, one often needs to recognize face images from a very long distance. Such recognition tasks are very challenging, since such images are typically with very low resolution (VLR). However, if one simply downsamples high-resolution (HR) training images for recognizing the VLR test inputs, or if one directly upsamples the VLR inputs for matching the HR training data, the resulting recognition performance would not be satisfactory. In this paper, we propose a joint face hallucination and recognition approach based on sparse representation. Given a VLR input image, our method is able to synthesize its person-specific HR version with recognition guarantees. In our experiments, we consider two different face image datasets. Empirical results will support the use of our approach for both VLR face recognition. In addition, compared to state-of-the-art super-resolution (SR) methods, we will also show that our method results in improved quality for the recovered HR face images.",
"Due to importance of security in the society, monitoring activities and recognizing specific people through surveillance video camera is playing an important role. One of the main issues in such activity rises from the fact that cameras do not meet the resolution requirement for many face recognition algorithm. In order to solve this issue, in this paper we are proposing a new system which super resolve the image using sparse representation with the specific dictionary involving many natural and facial images followed by Hidden Markov Model and Support vector machine based face recognition. The proposed system has been tested on many well-known face databases such as FERET, HeadPose, and Essex University databases as well as our recently introduced iCV Face Recognition database (iFRD). The experimental results shows that the recognition rate is increasing considerably after apply the super resolution by using facial and natural image dictionary.",
"In video surveillance, the captured face images are usually of low resolution (LR). Thus, a framework based on singular value decomposition (SVD) for performing both face hallucination and recognition simultaneously is proposed in this paper. Conventionally, LR face recognition is carried out by super-resolving the LR input face first, and then performing face recognition to identify the input face. By considering face hallucination and recognition simultaneously, the accuracy of both the hallucination and the recognition can be improved. In this paper, singular values are first proved to be effective for representing face images, and the singular values of a face image at different resolutions have approximately a linear relation. In our algorithm, each face image is represented using SVD. For each LR input face, the corresponding LR and high-resolution (HR) face-image pairs can then be selected from the face gallery. Based on these selected LR–HR pairs, the mapping functions for interpolating the two matrices in the SVD representation for the reconstruction of HR face images can be learned more accurately. Therefore, the final estimation of the high-frequency details of the HR face images will become more reliable and effective. The experimental results demonstrate that our proposed framework can achieve promising results for both face hallucination and recognition.",
"Despite the breakthroughs in accuracy and speed of single image super-resolution using faster and deeper convolutional neural networks, one central problem remains largely unsolved: how do we recover the finer texture details when we super-resolve at large upscaling factors? The behavior of optimization-based super-resolution methods is principally driven by the choice of the objective function. Recent work has largely focused on minimizing the mean squared reconstruction error. The resulting estimates have high peak signal-to-noise ratios, but they are often lacking high-frequency details and are perceptually unsatisfying in the sense that they fail to match the fidelity expected at the higher resolution. In this paper, we present SRGAN, a generative adversarial network (GAN) for image super-resolution (SR). To our knowledge, it is the first framework capable of inferring photo-realistic natural images for 4x upscaling factors. To achieve this, we propose a perceptual loss function which consists of an adversarial loss and a content loss. The adversarial loss pushes our solution to the natural image manifold using a discriminator network that is trained to differentiate between the super-resolved images and original photo-realistic images. In addition, we use a content loss motivated by perceptual similarity instead of similarity in pixel space. Our deep residual network is able to recover photo-realistic textures from heavily downsampled images on public benchmarks. An extensive mean-opinion-score (MOS) test shows hugely significant gains in perceptual quality using SRGAN. The MOS scores obtained with SRGAN are closer to those of the original high-resolution images than to those obtained with any state-of-the-art method.",
"We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low high-resolution images. The mapping is represented as a deep convolutional neural network (CNN) [15] that takes the low-resolution image as the input and outputs the high-resolution one. We further show that traditional sparse-coding-based SR methods can also be viewed as a deep convolutional network. But unlike traditional methods that handle each component separately, our method jointly optimizes all layers. Our deep CNN has a lightweight structure, yet demonstrates state-of-the-art restoration quality, and achieves fast speed for practical on-line usage."
]
}
|
1811.09998
|
2901505625
|
Typically, the deployment of face recognition models in the wild needs to identify low-resolution faces with extremely low computational cost. To address this problem, a feasible solution is compressing a complex face model to achieve higher speed and lower memory at the cost of minimal performance drop. Inspired by that, this paper proposes a learning approach to recognize low-resolution faces via selective knowledge distillation. In this approach, a two-stream convolutional neural network (CNN) is first initialized to recognize high-resolution faces and resolution-degraded faces with a teacher stream and a student stream, respectively. The teacher stream is represented by a complex CNN for high-accuracy recognition, and the student stream is represented by a much simpler CNN for low-complexity recognition. To avoid significant performance drop at the student stream, we then selectively distil the most informative facial features from the teacher stream by solving a sparse graph optimization problem, which are then used to regularize the fine-tuning process of the student stream. In this way, the student stream is actually trained by simultaneously handling two tasks with limited computational resources: approximating the most informative facial cues via feature regression, and recovering the missing facial cues via low-resolution face classification. Experimental results show that the student stream performs impressively in recognizing low-resolution faces and costs only 0.15-MB memory and runs at 418 faces per second on CPU and 9433 faces per second on GPU.
|
Instead of reconstructing high-resolution faces, a more direct approach is embedding low-resolution faces into various external contexts to recover the missing features during resolution degradation. Inspired by that, some approaches proposed transforming both high-resolution and low-resolution faces into a unified feature space for matching @cite_3 @cite_21 @cite_45 @cite_56 @cite_60 @cite_4 @cite_31 , while in @cite_35 @cite_49 the multi-scale (multi-resolution) faces were simultaneously analyzed to extract better features. @cite_25 , the multidimensional scaling was adopted to learn a common transformation matrix to simultaneously transform the facial features of low-resolution and high-resolution training images. Shekhar al @cite_14 proposed a joint sparse coding technique for robust recognition at low-resolution, while Wang al @cite_67 attempted to solve very low resolution recognition problem using deep learning methods. In @cite_39 , CNNs were adopted with a manifold-based track comparison strategy for low-resolution face recognition in videos.
|
{
"cite_N": [
"@cite_35",
"@cite_67",
"@cite_14",
"@cite_4",
"@cite_60",
"@cite_21",
"@cite_3",
"@cite_56",
"@cite_39",
"@cite_45",
"@cite_49",
"@cite_31",
"@cite_25"
],
"mid": [
"2213726222",
"2963102887",
"2735525189",
"1967482855",
"2275479513",
"1747367457",
"2054515210",
"2345296463",
"2554173462",
"1889464797",
"2093422693",
"2726947518",
"2332698925"
],
"abstract": [
"In real world person re-identification (re-id), images of people captured at very different resolutions from different locations need be matched. Existing re-id models typically normalise all person images to the same size. However, a low-resolution (LR) image contains much less information about a person, and direct image scaling and simple size normalisation as done in conventional re-id methods cannot compensate for the loss of information. To solve this LR person re-id problem, we propose a novel joint multi-scale learning framework, termed joint multi-scale discriminant component analysis (JUDEA). The key component of this framework is a heterogeneous class mean discrepancy (HCMD) criterion for cross-scale image domain alignment, which is optimised simultaneously with discriminant modelling across multiple scales in the joint learning framework. Our experiments show that the proposed JUDEA framework outperforms existing representative re-id methods as well as other related LR visual matching models applied for the LR person re-id problem.",
"Visual recognition research often assumes a sufficient resolution of the region of interest (ROI). That is usually violated in practice, inspiring us to explore the Very Low Resolution Recognition (VLRR) problem. Typically, the ROI in a VLRR problem can be smaller than 16 16 pixels, and is challenging to be recognized even by human experts. We attempt to solve the VLRR problem using deep learning methods. Taking advantage of techniques primarily in super resolution, domain adaptation and robust regression, we formulate a dedicated deep learning method and demonstrate how these techniques are incorporated step by step. Any extra complexity, when introduced, is fully justified by both analysis and simulation results. The resulting Robust Partially Coupled Networks achieves feature enhancement and recognition simultaneously. It allows for both the flexibility to combat the LR-HR domain mismatch, and the robustness to outliers. Finally, the effectiveness of the proposed models is evaluated on three different VLRR tasks, including face identification, digit recognition and font recognition, all of which obtain very impressive performances.",
"Recognition of low resolution face images is a challenging problem in many practical face recognition systems. Methods have been proposed in the face recognition literature for the problem which assume that the probe is low resolution, but a high resolution gallery is available for recognition. These attempts have been aimed at modifying the probe image such that the resultant image provides better discrimination. We formulate the problem differently by leveraging the information available in the high resolution gallery image and propose a dictionary learning approach for classifying the low-resolution probe image. An important feature of our algorithm is that it can handle resolution change along with illumination variations. Furthermore, we also kernelize the algorithm to handle non-linearity in data and present a joint dictionary learning technique for robust recognition at low resolutions. The effectiveness of the proposed method is demonstrated using standard datasets and a challenging outdoor face dataset. It is shown that our method is efficient and can perform significantly better than many competitive low resolution face recognition algorithms.",
"In this letter, we propose a novel approach for learning coupled mappings to improve the performance of low-resolution (LR) face image recognition. The coupled mappings aim to project the LR probe images and high-resolution (HR) gallery images into a unified latent subspace, which is efficient to measure the similarity of face images with different resolutions. In the training phase, we first construct local optimization for each training sample according to the relationship of neighboring data points. The local optimization aims to: (1) ensure the consistency for each LR face image and corresponding HR one; (2) model the intrinsic geometric structure between each given sample and its neighbors; and (3) preserve the discriminative information across different subjects. We finally incorporate the local optimizations together for building the global structure. The coupled mappings can be learned by solving a standard eigen-decomposition problem, which avoids the small-sample-size problem. Experimental results demonstrate the effectiveness of the proposed method on public face databases.",
"This brief paper presents a novel method for low-resolution face recognition. We introduce a generalized bipartite graph to discretely approximate the underlying manifold structure of face sets with different resolutions. Unlike traditional graph-based methods that only construct the graph based on one sample set, the proposed method constructs the generalized bipartite graph on two heterogeneous sample sets and contains more completed information. Our method learns a couple of mappings that project the face sets with different dimensions into a unified feature space which favors the task of classification. Specifically, in this unified space, our method preserves within-class local geometrical structure according to the network topology of the generalized bipartite graph and maximizes between-class separability at the same time. Experimental results on two benchmark face databases demonstrate the effectiveness of our proposed algorithm. HighlightsWe propose a novel method for low-resolution face recognition.We introduce the generalized bipartite graph to approximate the manifold structure.Our method preserves within-class local geometrical structure.Our method maximizes between-class separability.Low and high resolution faces are projected to a unified discriminative subspace.",
"Abstract This study develops a novel efficient coupled distance metric learning algorithm called coupled marginal discriminant mappings (CMDM). It can provide a low-resolution face recognition method with coupled mappings, making the data points in the original high and low resolution features projected into a unified space, where classification is implemented. At the same time, it makes samples from the same class gather more closely while makes samples of distinct class disperse more separately with a large margin. Thus, for the low-resolution face recognition issue, the proposed method can be leveraged to avoid dimensional mismatch problem and fill different-resolution data gap. The experimental evaluation based on the AR and FERET face databases demonstrates that CMDM can achieve highly competitive performance comparing favorably with the existing state-of-the-art low-resolution face recognition methods.",
"This paper addresses the very low resolution (VLR) problem in face recognition in which the resolution of the face image to be recognized is lower than 16 × 16. With the increasing demand of surveillance camera-based applications, the VLR problem happens in many face application systems. Existing face recognition algorithms are not able to give satisfactory performance on the VLR face image. While face super-resolution (SR) methods can be employed to enhance the resolution of the images, the existing learning-based face SR methods do not perform well on such a VLR face image. To overcome this problem, this paper proposes a novel approach to learn the relationship between the high-resolution image space and the VLR image space for face SR. Based on this new approach, two constraints, namely, new data and discriminative constraints, are designed for good visuality and face recognition applications under the VLR problem, respectively. Experimental results show that the proposed SR algorithm based on relationship learning outperforms the existing algorithms in public face databases.",
"Most face recognition techniques have been successful in dealing with high-resolution (HR) frontal face images. However, real-world face recognition systems are often confronted with the low-resolution (LR) face images with pose and illumination variations. This is a very challenging issue, especially under the constraint of using only a single gallery image per person. To address the problem, we propose a novel approach called coupled kernel-based enhanced discriminant analysis (CKEDA). CKEDA aims to simultaneously project the features from LR non-frontal probe images and HR frontal gallery ones into a common space where discrimination property is maximized. There are four advantages of the proposed approach: 1) by using the appropriate kernel function, the data becomes linearly separable, which is beneficial for recognition; 2) inspired by linear discriminant analysis (LDA), we integrate multiple discriminant factors into our objective function to enhance the discrimination property; 3) we use the gallery extended trick to improve the recognition performance for a single gallery image per person problem; 4) our approach can address the problem of matching LR non-frontal probe images with HR frontal gallery images, which is difficult for most existing face recognition techniques. Experimental evaluation on the multi-PIE dataset signifies highly competitive performance of our algorithm.",
"Security and safety applications such as surveillance or forensics demand face recognition in low-resolution video data. We propose a face recognition method based on a Convolutional Neural Network (CNN) with a manifold-based track comparison strategy for low-resolution video face recognition. The low-resolution domain is addressed by adjusting the network architecture to prevent bottlenecks or significant upscaling of face images. The CNN is trained with a combination of a large-scale self-collected video face dataset and large-scale public image face datasets resulting in about 1.4M training images. To handle large amounts of video data and for effective comparison, the CNN face descriptors are compared efficiently on track level by local patch means. Our setup achieves 80.3 percent accuracy on a 32×32 pixels low-resolution version of the YouTube Faces Database and outperforms local image descriptors as well as the state-of-the-art VGG-Face network [20] in this domain. The superior performance of the proposed method is confirmed on a self-collected in-the-wild surveillance dataset.",
"Abstract Face images captured by surveillance cameras usually have low-resolution (LR) in addition to uncontrolled poses and illumination conditions, all of which adversely affect the performance of face matching algorithms. In this paper, we develop a novel method to address the problem of matching a LR or poor quality face image to a gallery of high-resolution (HR) face images. In recent years, extensive efforts have been made on LR face recognition (FR) research. Previous research has focused on introducing a learning based super-resolution (LBSR) method before matching or transforming LR and HR faces into a unified feature space (UFS) for matching. To identify LR faces, we present a method called coupled discriminant multi-manifold analysis (CDMMA). In CDMMA, we first explore the neighborhood information as well as local geometric structure of the multi-manifold space spanned by the samples. And then, we explicitly learn two mappings to project LR and HR faces to a unified discriminative feature space (UDFS) through a supervised manner, where the discriminative information is maximized for classification. After that, the conventional classification method is applied in the CDMMA for final identification. Experimental results conducted on two standard face recognition databases demonstrate the superiority of the proposed CDMMA.",
"For face recognition, image features are first extracted and then matched to those features in a gallery set. The amount of information and the effectiveness of the features used will determine the recognition performance. In this paper, we propose a novel face recognition approach using information about face images at higher and lower resolutions so as to enhance the information content of the features that are extracted and combined at different resolutions. As the features from different resolutions should closely correlate with each other, we employ the cascaded generalized canonical correlation analysis (GCCA) to fuse the information to form a single feature vector for face recognition. To improve the performance and efficiency, we also employ \"Gabor-feature hallucination\", which predicts the high-resolution (HR) Gabor features from the Gabor features of a face image directly by local linear regression. We also extend the algorithm to low-resolution (LR) face recognition, in which the medium-resolution (MR) and HR Gabor features of a LR input image are estimated directly. The LR Gabor features and the predicted MR and HR Gabor features are then fused using GCCA for LR face recognition. Our algorithm can avoid having to perform the interpolation super-resolution of face images and having to extract HR Gabor features. Experimental results show that the proposed methods have a superior recognition rate and are more efficient than traditional methods. A face recognition approach which combines images at different resolutions is proposed.A low-resolution face recognition algorithm based on fusing images at different resolutions is proposed.A method for feature hallucination is proposed.",
"Due to large distances between surveillance cameras and subjects, the captured images usually have low resolution in addition to uncontrolled poses and illumination conditions that adversely affect the performance of face recognition algorithms. In this paper, we present a low-resolution face recognition technique based on Discriminant Correlation Analysis (DCA). DCA analyzes the correlation of the features in high-resolution and low-resolution images and aims to find projections that maximize the pair-wise correlations between the two feature sets and at the same time, separate the classes within each set. This makes it possible to project the features extracted from high-resolution and low-resolution images into a common space, in which we can apply matching. The proposed method is computationally efficient and can be applied to challenging real-time applications such as recognition of several faces appearing in a crowded frame of a surveillance video. Extensive experiments performed on low-resolution surveillance images from the SCface database as well as FRGC database demonstrated the efficacy of our proposed approach in the recognition of low-resolution face images, which outperformed other state-of-the-art techniques.",
"We propose a completely automatic approach for recognizing low resolution face images captured in uncontrolled environment. The approach uses multidimensional scaling to learn a common transformation matrix for the entire face which simultaneously transforms the facial features of the low resolution and the high resolution training images such that the distance between them approximates the distance had both the images been captured under the same controlled imaging conditions. Stereo matching cost is used to obtain the similarity of two images in the transformed space. Though this gives very good recognition performance, the time taken for computing the stereo matching cost is significant. To overcome this limitation, we propose a reference-based approach in which each face image is represented by its stereo matching cost from a few reference images. Experimental evaluation on the real world challenging databases and comparison with the state-of-the-art super-resolution, classifier based and cross modal synthesis techniques show the effectiveness of the proposed algorithm."
]
}
|
1811.09975
|
2949691889
|
Variational autoencoders were proven successful in domains such as computer vision and speech processing. Their adoption for modeling user preferences is still unexplored, although recently it is starting to gain attention in the current literature. In this work, we propose a model which extends variational autoencoders by exploiting the rich information present in the past preference history. We introduce a recurrent version of the VAE, where instead of passing a subset of the whole history regardless of temporal dependencies, we rather pass the consumption sequence subset through a recurrent neural network. At each time-step of the RNN, the sequence is fed through a series of fully-connected layers, the output of which models the probability distribution of the most likely future preferences. We show that handling temporal information is crucial for improving the accuracy of the VAE: In fact, our model beats the current state-of-the-art by valuable margins because of its ability to capture temporal dependencies among the user-consumption sequence using the recurrent encoder still keeping the fundamentals of variational autoencoders intact.
|
Most approaches disregard the temporal order of the preferences in a user's history. Among these, latent variable models @cite_43 @cite_19 @cite_17 @cite_7 @cite_25 @cite_50 @cite_24 @cite_10 @cite_42 @cite_6 @cite_38 were proven extremely effective in modeling user preferences and providing reliable recommendations. Essentially, these approaches embed users and items into latent spaces that translate relatedness into geometrical closeness. The latent embeddings can be used to decompose the large sparse @cite_17 @cite_19 @cite_25 , to devise item similarity @cite_50 @cite_37 , or more generally to parameterize probability distributions for item preference @cite_7 @cite_12 @cite_24 @cite_10 and sharpen the prediction quality by means of meaningful priors.
|
{
"cite_N": [
"@cite_38",
"@cite_37",
"@cite_7",
"@cite_42",
"@cite_6",
"@cite_24",
"@cite_19",
"@cite_43",
"@cite_50",
"@cite_10",
"@cite_25",
"@cite_12",
"@cite_17"
],
"mid": [
"2280165275",
"",
"2950975304",
"2108920354",
"",
"2135790056",
"2137245235",
"2049455633",
"1987431925",
"2094286023",
"2135505871",
"",
"2085040216"
],
"abstract": [
"In this work we perform an analysis of probabilistic approaches to recommendation upon a different validation perspective, which focuses on accuracy metrics such as recall and precision of the recommendation list. Traditionally, state-of-art approches to recommendations consider the recommendation process from a \"missing value prediction\" perspective. This approach simplifies the model validation phase that is based on the minimization of standard error metrics such as RMSE. However, recent studies have pointed several limitations of this approach, showing that a lower RMSE does not necessarily imply improvements in terms of specific recommendations. We demonstrate that the underlying probabilistic framework offers several advantages over traditional methods, in terms of flexibility in the generation of the recommendation list and consequently in the accuracy of recommendation.",
"",
"Item recommendation is the task of predicting a personalized ranking on a set of items (e.g. websites, movies, products). In this paper, we investigate the most common scenario with implicit feedback (e.g. clicks, purchases). There are many methods for item recommendation from implicit feedback like matrix factorization (MF) or adaptive knearest-neighbor (kNN). Even though these methods are designed for the item prediction task of personalized ranking, none of them is directly optimized for ranking. In this paper we present a generic optimization criterion BPR-Opt for personalized ranking that is the maximum posterior estimator derived from a Bayesian analysis of the problem. We also provide a generic learning algorithm for optimizing models with respect to BPR-Opt. The learning method is based on stochastic gradient descent with bootstrap sampling. We show how to apply our method to two state-of-the-art recommender models: matrix factorization and adaptive kNN. Our experiments indicate that for the task of personalized ranking our optimization method outperforms the standard learning techniques for MF and kNN. The results show the importance of optimizing models for the right criterion.",
"The effectiveness of existing top-N recommendation methods decreases as the sparsity of the datasets increases. To alleviate this problem, we present an item-based method for generating top-N recommendations that learns the item-item similarity matrix as the product of two low dimensional latent factor matrices. These matrices are learned using a structural equation modeling approach, wherein the value being estimated is not used for its own estimation. A comprehensive set of experiments on multiple datasets at three different sparsity levels indicate that the proposed methods can handle sparse datasets effectively and outperforms other state-of-the-art top-N recommendation methods. The experimental results also show that the relative performance gains compared to competing methods increase as the data gets sparser.",
"",
"Researchers have access to large online archives of scientific articles. As a consequence, finding relevant papers has become more difficult. Newly formed online communities of researchers sharing citations provides a new way to solve this problem. In this paper, we develop an algorithm to recommend scientific articles to users of an online community. Our approach combines the merits of traditional collaborative filtering and probabilistic topic modeling. It provides an interpretable latent structure for users and items, and can form recommendations about both existing and newly published articles. We study a large subset of data from CiteULike, a bibliography sharing service, and show that our algorithm provides a more effective recommender system than traditional collaborative filtering.",
"Many existing approaches to collaborative filtering can neither handle very large datasets nor easily deal with users who have very few ratings. In this paper we present the Probabilistic Matrix Factorization (PMF) model which scales linearly with the number of observations and, more importantly, performs well on the large, sparse, and very imbalanced Netflix dataset. We further extend the PMF model to include an adaptive prior on the model parameters and show how the model capacity can be controlled automatically. Finally, we introduce a constrained version of the PMF model that is based on the assumption that users who have rated similar sets of movies are likely to have similar preferences. The resulting model is able to generalize considerably better for users with very few ratings. When the predictions of multiple PMF models are linearly combined with the predictions of Restricted Boltzmann Machines models, we achieve an error rate of 0.8861, that is nearly 7 better than the score of Netflix's own system.",
"Collaborative filtering aims at learning predictive models of user preferences, interests or behavior from community data, that is, a database of available user preferences. In this article, we describe a new family of model-based algorithms designed for this task. These algorithms rely on a statistical modelling technique that introduces latent class variables in a mixture model setting to discover user communities and prototypical interest profiles. We investigate several variations to deal with discrete and continuous response variables as well as with different objective functions. The main advantages of this technique over standard memory-based methods are higher accuracy, constant time prediction, and an explicit and compact model representation. The latter can also be used to mine for user communitites. The experimental evaluation shows that substantial improvements in accucracy over existing methods and published results can be obtained.",
"This paper focuses on developing effective and efficient algorithms for top-N recommender systems. A novel Sparse Linear Method (SLIM) is proposed, which generates top-N recommendations by aggregating from user purchase rating profiles. A sparse aggregation coefficient matrix W is learned from SLIM by solving an 1-norm and 2-norm regularized optimization problem. W is demonstrated to produce high quality recommendations and its sparsity allows SLIM to generate recommendations very fast. A comprehensive set of experiments is conducted by comparing the SLIM method and other state-of-the-art top-N recommendation methods. The experiments show that SLIM achieves significant improvements both in run time performance and recommendation quality over the best existing methods.",
"Factorization approaches provide high accuracy in several important prediction problems, for example, recommender systems. However, applying factorization approaches to a new prediction problem is a nontrivial task and requires a lot of expert knowledge. Typically, a new model is developed, a learning algorithm is derived, and the approach has to be implemented. Factorization machines (FM) are a generic approach since they can mimic most factorization models just by feature engineering. This way, factorization machines combine the generality of feature engineering with the superiority of factorization models in estimating interactions between categorical variables of large domain. libFM is a software implementation for factorization machines that features stochastic gradient descent (SGD) and alternating least-squares (ALS) optimization, as well as Bayesian inference using Markov Chain Monto Carlo (MCMC). This article summarizes the recent research on factorization machines both in terms of modeling and learning, provides extensions for the ALS and MCMC algorithms, and describes the software tool libFM .",
"We propose fLDA, a novel matrix factorization method to predict ratings in recommender system applications where a \"bag-of-words\" representation for item meta-data is natural. Such scenarios are commonplace in web applications like content recommendation, ad targeting and web search where items are articles, ads and web pages respectively. Because of data sparseness, regularization is key to good predictive accuracy. Our method works by regularizing both user and item factors simultaneously through user features and the bag of words associated with each item. Specifically, each word in an item is associated with a discrete latent factor often referred to as the topic of the word; item topics are obtained by averaging topics across all words in an item. Then, user rating on an item is modeled as user's affinity to the item's topics where user affinity to topics (user factors) and topic assignments to words in items (item factors) are learned jointly in a supervised fashion. To avoid overfitting, user and item factors are regularized through Gaussian linear regression and Latent Dirichlet Allocation (LDA) priors respectively. We show our model is accurate, interpretable and handles both cold-start and warm-start scenarios seamlessly through a single model. The efficacy of our method is illustrated on benchmark datasets and a new dataset from Yahoo! Buzz where fLDA provides superior predictive accuracy in cold-start scenarios and is comparable to state-of-the-art methods in warm-start scenarios. As a by-product, fLDA also identifies interesting topics that explains user-item interactions. Our method also generalizes a recently proposed technique called supervised LDA (sLDA) to collaborative filtering applications. While sLDA estimates item topic vectors in a supervised fashion for a single regression, fLDA incorporates multiple regressions (one for each user) in estimating the item factors.",
"",
"Low-rank matrix approximation methods provide one of the simplest and most effective approaches to collaborative filtering. Such models are usually fitted to data by finding a MAP estimate of the model parameters, a procedure that can be performed efficiently even on very large datasets. However, unless the regularization parameters are tuned carefully, this approach is prone to overfitting because it finds a single point estimate of the parameters. In this paper we present a fully Bayesian treatment of the Probabilistic Matrix Factorization (PMF) model in which model capacity is controlled automatically by integrating over all model parameters and hyperparameters. We show that Bayesian PMF models can be efficiently trained using Markov chain Monte Carlo methods by applying them to the Netflix dataset, which consists of over 100 million movie ratings. The resulting models achieve significantly higher prediction accuracy than PMF models trained using MAP estimation."
]
}
|
1811.09975
|
2949691889
|
Variational autoencoders were proven successful in domains such as computer vision and speech processing. Their adoption for modeling user preferences is still unexplored, although recently it is starting to gain attention in the current literature. In this work, we propose a model which extends variational autoencoders by exploiting the rich information present in the past preference history. We introduce a recurrent version of the VAE, where instead of passing a subset of the whole history regardless of temporal dependencies, we rather pass the consumption sequence subset through a recurrent neural network. At each time-step of the RNN, the sequence is fed through a series of fully-connected layers, the output of which models the probability distribution of the most likely future preferences. We show that handling temporal information is crucial for improving the accuracy of the VAE: In fact, our model beats the current state-of-the-art by valuable margins because of its ability to capture temporal dependencies among the user-consumption sequence using the recurrent encoder still keeping the fundamentals of variational autoencoders intact.
|
The recent literature is currently focusing on deep learning, which shows substantial advantages over traditional approaches. For example, (NCF) @cite_46 generalizes matrix factorization to a non-linear setting, where users, items and preferences are modeled through a simple multilayer perceptron network that exploits latent factor transformations.
|
{
"cite_N": [
"@cite_46"
],
"mid": [
"2951707557"
],
"abstract": [
"In recent years, deep neural networks have yielded immense success on speech recognition, computer vision and natural language processing. However, the exploration of deep neural networks on recommender systems has received relatively less scrutiny. In this work, we strive to develop techniques based on neural networks to tackle the key problem in recommendation -- collaborative filtering -- on the basis of implicit feedback. Although some recent work has employed deep learning for recommendation, they primarily used it to model auxiliary information, such as textual descriptions of items and acoustic features of musics. When it comes to model the key factor in collaborative filtering -- the interaction between user and item features, they still resorted to matrix factorization and applied an inner product on the latent features of users and items. By replacing the inner product with a neural architecture that can learn an arbitrary function from data, we present a general framework named NCF, short for Neural network-based Collaborative Filtering. NCF is generic and can express and generalize matrix factorization under its framework. To supercharge NCF modelling with non-linearities, we propose to leverage a multi-layer perceptron to learn the user-item interaction function. Extensive experiments on two real-world datasets show significant improvements of our proposed NCF framework over the state-of-the-art methods. Empirical evidence shows that using deeper layers of neural networks offers better recommendation performance."
]
}
|
1811.09975
|
2949691889
|
Variational autoencoders were proven successful in domains such as computer vision and speech processing. Their adoption for modeling user preferences is still unexplored, although recently it is starting to gain attention in the current literature. In this work, we propose a model which extends variational autoencoders by exploiting the rich information present in the past preference history. We introduce a recurrent version of the VAE, where instead of passing a subset of the whole history regardless of temporal dependencies, we rather pass the consumption sequence subset through a recurrent neural network. At each time-step of the RNN, the sequence is fed through a series of fully-connected layers, the output of which models the probability distribution of the most likely future preferences. We show that handling temporal information is crucial for improving the accuracy of the VAE: In fact, our model beats the current state-of-the-art by valuable margins because of its ability to capture temporal dependencies among the user-consumption sequence using the recurrent encoder still keeping the fundamentals of variational autoencoders intact.
|
Notably, prominent deep learning approaches to collaborative filtering are based on the idea of autoencoding the features from the preference matrix. @cite_44 exploits autoencoders to encode preference histories. Unseen preferences can be devised by looking at the reconstructed decoding, which is shaped to include scores for all possible items of interest. Autoencoders are also amenable to consider side information @cite_9 to mitigate the sparsity of the data and to tackle the cold start problem.
|
{
"cite_N": [
"@cite_44",
"@cite_9"
],
"mid": [
"1720514416",
"2615395371"
],
"abstract": [
"This paper proposes AutoRec, a novel autoencoder framework for collaborative filtering (CF). Empirically, AutoRec's compact and efficiently trainable model outperforms state-of-the-art CF techniques (biased matrix factorization, RBM-CF and LLORMA) on the Movielens and Netflix datasets.",
"Neural networks have not been widely studied in Collaborative Filtering. For instance, no paper using neural networks was published during the Net-flix Prize apart from 's work on Restricted Boltzmann Machine (RBM) [14]. While deep learning has tremendous success in image and speech recognition, sparse inputs received less attention and remains a challenging problem for neural networks. Nonetheless, sparse inputs are critical for collaborative filtering. In this paper, we introduce a neural network architecture which computes a non-linear matrix factorization from sparse rating inputs. We show experimentally on the movieLens and jester dataset that our method performs as well as the best collaborative filtering algorithms. We provide an implementation of the algorithm as a reusable plugin for Torch [4], a popular neural network framework."
]
}
|
1811.09975
|
2949691889
|
Variational autoencoders were proven successful in domains such as computer vision and speech processing. Their adoption for modeling user preferences is still unexplored, although recently it is starting to gain attention in the current literature. In this work, we propose a model which extends variational autoencoders by exploiting the rich information present in the past preference history. We introduce a recurrent version of the VAE, where instead of passing a subset of the whole history regardless of temporal dependencies, we rather pass the consumption sequence subset through a recurrent neural network. At each time-step of the RNN, the sequence is fed through a series of fully-connected layers, the output of which models the probability distribution of the most likely future preferences. We show that handling temporal information is crucial for improving the accuracy of the VAE: In fact, our model beats the current state-of-the-art by valuable margins because of its ability to capture temporal dependencies among the user-consumption sequence using the recurrent encoder still keeping the fundamentals of variational autoencoders intact.
|
Hybrid approaches that integrate latent variable modeling and deep learning have also gained attention. @cite_52 embeds a @cite_53 into a Bayesian matrix factorization setting. Similarly, @cite_28 exploits the notion of @cite_55 to learn latent item representations that are integrated into the model @cite_1 .
|
{
"cite_N": [
"@cite_28",
"@cite_55",
"@cite_53",
"@cite_1",
"@cite_52"
],
"mid": [
"2613725451",
"",
"2145094598",
"1994389483",
"2950316093"
],
"abstract": [
"Collaborative filtering (CF) has been successfully used to provide users with personalized products and services. However, dealing with the increasing sparseness of user-item matrix still remains a challenge. To tackle such issue, hybrid CF such as combining with content based filtering and leveraging side information of users and items has been extensively studied to enhance performance. However, most of these approaches depend on hand-crafted feature engineering, which is usually noise-prone and biased by different feature extraction and selection schemes. In this paper, we propose a new hybrid model by generalizing contractive auto-encoder paradigm into matrix factorization framework with good scalability and computational efficiency, which jointly models content information as representations of effectiveness and compactness, and leverage implicit user feedback to make accurate recommendations. Extensive experiments conducted over three large-scale real datasets indicate the proposed approach outperforms the compared methods for item recommendation.",
"",
"We explore an original strategy for building deep networks, based on stacking layers of denoising autoencoders which are trained locally to denoise corrupted versions of their inputs. The resulting algorithm is a straightforward variation on the stacking of ordinary autoencoders. It is however shown on a benchmark of classification problems to yield significantly lower classification error, thus bridging the performance gap with deep belief networks (DBN), and in several cases surpassing it. Higher level representations learnt in this purely unsupervised fashion also help boost the performance of subsequent SVM classifiers. Qualitative experiments show that, contrary to ordinary autoencoders, denoising autoencoders are able to learn Gabor-like edge detectors from natural image patches and larger stroke detectors from digit images. This work clearly establishes the value of using a denoising criterion as a tractable unsupervised objective to guide the learning of useful higher level representations.",
"Recommender systems provide users with personalized suggestions for products or services. These systems often rely on Collaborating Filtering (CF), where past transactions are analyzed in order to establish connections between users and products. The two more successful approaches to CF are latent factor models, which directly profile both users and products, and neighborhood models, which analyze similarities between products or users. In this work we introduce some innovations to both approaches. The factor and neighborhood models can now be smoothly merged, thereby building a more accurate combined model. Further accuracy improvements are achieved by extending the models to exploit both explicit and implicit feedback by the users. The methods are tested on the Netflix data. Results are better than those previously published on that dataset. In addition, we suggest a new evaluation metric, which highlights the differences among methods, based on their performance at a top-K recommendation task.",
"Collaborative filtering (CF) is a successful approach commonly used by many recommender systems. Conventional CF-based methods use the ratings given to items by users as the sole source of information for learning to make recommendation. However, the ratings are often very sparse in many applications, causing CF-based methods to degrade significantly in their recommendation performance. To address this sparsity problem, auxiliary information such as item content information may be utilized. Collaborative topic regression (CTR) is an appealing recent method taking this approach which tightly couples the two components that learn from two different sources of information. Nevertheless, the latent representation learned by CTR may not be very effective when the auxiliary information is very sparse. To address this problem, we generalize recent advances in deep learning from i.i.d. input to non-i.i.d. (CF-based) input and propose in this paper a hierarchical Bayesian model called collaborative deep learning (CDL), which jointly performs deep representation learning for the content information and collaborative filtering for the ratings (feedback) matrix. Extensive experiments on three real-world datasets from different domains show that CDL can significantly advance the state of the art."
]
}
|
1811.09975
|
2949691889
|
Variational autoencoders were proven successful in domains such as computer vision and speech processing. Their adoption for modeling user preferences is still unexplored, although recently it is starting to gain attention in the current literature. In this work, we propose a model which extends variational autoencoders by exploiting the rich information present in the past preference history. We introduce a recurrent version of the VAE, where instead of passing a subset of the whole history regardless of temporal dependencies, we rather pass the consumption sequence subset through a recurrent neural network. At each time-step of the RNN, the sequence is fed through a series of fully-connected layers, the output of which models the probability distribution of the most likely future preferences. We show that handling temporal information is crucial for improving the accuracy of the VAE: In fact, our model beats the current state-of-the-art by valuable margins because of its ability to capture temporal dependencies among the user-consumption sequence using the recurrent encoder still keeping the fundamentals of variational autoencoders intact.
|
The introduction of the variational autoencoding framework @cite_3 @cite_54 has suggested a tighter coupling between deep learning and latent variable modeling. (CVA) @cite_4 and (HVAE) @cite_5 exploit side information to feed a variational autoencoder whose goal is to produce a latent representation of the items. In CVA, the preference matrix is hence modeled by combining user and item embeddings with the item latent representations, while HVAE uses another variational autoencoder to reproduce the whole users' preference history. By contrast, @cite_45 proposes a neural generative model where a user's history is modeled through a multinomial likelihood conditioned to a latent user representation which in turn is modeled through a variational autoencoder.
|
{
"cite_N": [
"@cite_4",
"@cite_54",
"@cite_3",
"@cite_45",
"@cite_5"
],
"mid": [
"2725606191",
"1909320841",
"",
"2787512446",
"2885456372"
],
"abstract": [
"Modern recommender systems usually employ collaborative filtering with rating information to recommend items to users due to its successful performance. However, because of the drawbacks of collaborative-based methods such as sparsity, cold start, etc., more attention has been drawn to hybrid methods that consider both the rating and content information. Most of the previous works in this area cannot learn a good representation from content for recommendation task or consider only text modality of the content, thus their methods are very limited in current multimedia scenario. This paper proposes a Bayesian generative model called collaborative variational autoencoder (CVAE) that considers both rating and content for recommendation in multimedia scenario. The model learns deep latent representations from content data in an unsupervised manner and also learns implicit relationships between items and users from both content and rating. Unlike previous works with denoising criteria, the proposed CVAE learns a latent distribution for content in latent space instead of observation space through an inference network and can be easily extended to other multimedia modalities other than text. Experiments show that CVAE is able to significantly outperform the state-of-the-art recommendation methods with more robust performance.",
"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 approximate posterior distributions, and that acts as a stochastic encoder of the data. We develop stochastic back-propagation -- rules for back-propagation through stochastic variables -- and use this to develop an algorithm that allows for joint optimisation of the parameters of both the generative and recognition model. We demonstrate on several real-world data sets that the model generates realistic samples, provides accurate imputations of missing data and is a useful tool for high-dimensional data visualisation.",
"",
"We extend variational autoencoders (VAEs) to collaborative filtering for implicit feedback. This non-linear probabilistic model enables us to go beyond the limited modeling capacity of linear factor models which still largely dominate collaborative filtering research.We introduce a generative model with multinomial likelihood and use Bayesian inference for parameter estimation. Despite widespread use in language modeling and economics, the multinomial likelihood receives less attention in the recommender systems literature. We introduce a different regularization parameter for the learning objective, which proves to be crucial for achieving competitive performance. Remarkably, there is an efficient way to tune the parameter using annealing. The resulting model and learning algorithm has information-theoretic connections to maximum entropy discrimination and the information bottleneck principle. Empirically, we show that the proposed approach significantly outperforms several state-of-the-art baselines, including two recently-proposed neural network approaches, on several real-world datasets. We also provide extended experiments comparing the multinomial likelihood with other commonly used likelihood functions in the latent factor collaborative filtering literature and show favorable results. Finally, we identify the pros and cons of employing a principled Bayesian inference approach and characterize settings where it provides the most significant improvements.",
"In today's day and age when almost every industry has an online presence with users interacting in online marketplaces, personalized recommendations have become quite important. Traditionally, the problem of collaborative filtering has been tackled using Matrix Factorization which is linear in nature. We extend the work of [11] on using variational autoencoders (VAEs) for collaborative filtering with implicit feedback by proposing a hybrid, multi-modal approach. Our approach combines movie embeddings (learned from a sibling VAE network) with user ratings from the Movielens 20M dataset and applies it to the task of movie recommendation. We empirically show how the VAE network is empowered by incorporating movie embeddings. We also visualize movie and user embeddings by clustering their latent representations obtained from a VAE."
]
}
|
1811.09975
|
2949691889
|
Variational autoencoders were proven successful in domains such as computer vision and speech processing. Their adoption for modeling user preferences is still unexplored, although recently it is starting to gain attention in the current literature. In this work, we propose a model which extends variational autoencoders by exploiting the rich information present in the past preference history. We introduce a recurrent version of the VAE, where instead of passing a subset of the whole history regardless of temporal dependencies, we rather pass the consumption sequence subset through a recurrent neural network. At each time-step of the RNN, the sequence is fed through a series of fully-connected layers, the output of which models the probability distribution of the most likely future preferences. We show that handling temporal information is crucial for improving the accuracy of the VAE: In fact, our model beats the current state-of-the-art by valuable margins because of its ability to capture temporal dependencies among the user-consumption sequence using the recurrent encoder still keeping the fundamentals of variational autoencoders intact.
|
Within the context of collaborative filtering, a strong effort has also been made to model temporal dynamics within the history of user preferences @cite_34 . The model (FPMC) @cite_36 , for example, proposes a combination of matrix factorization and Markov chains. FPMC considers personalized first-order transition probabilities between items, which are modeled by decomposing the underlying tensor through user and item embeddings. Transition probabilities can also be measured by exploiting more sophisticated modeling @cite_15 @cite_26 , where users are mapped into translation (latent) vectors operating on item sequences and consequently a transition corresponds to a geometric affinity of these latent vectors. Orthogonally, Markov dependencies can also be exploited to model dependencies between latent variables @cite_49 , thus resulting in richer formalizations and more accurate recommendations.
|
{
"cite_N": [
"@cite_26",
"@cite_36",
"@cite_49",
"@cite_15",
"@cite_34"
],
"mid": [
"2734755249",
"2171279286",
"2013676140",
"2524638710",
""
],
"abstract": [
"Modeling the complex interactions between users and items as well as amongst items themselves is at the core of designing successful recommender systems. One classical setting is predicting users' personalized sequential behavior (or 'next-item' recommendation), where the challenges mainly lie in modeling 'third-order' interactions between a user, her previously visited item(s), and the next item to consume. Existing methods typically decompose these higher-order interactions into a combination of pairwise relationships, by way of which user preferences (user-item interactions) and sequential patterns (item-item interactions) are captured by separate components. In this paper, we propose a unified method, TransRec, to model such third-order relationships for large-scale sequential prediction. Methodologically, we embed items into a 'transition space' where users are modeled as translation vectors operating on item sequences. Empirically, this approach outperforms the state-of-the-art on a wide spectrum of real-world datasets. Data and code are available at https: sites.google.com a eng.ucsd.edu ruining-he .",
"Recommender systems are an important component of many websites. Two of the most popular approaches are based on matrix factorization (MF) and Markov chains (MC). MF methods learn the general taste of a user by factorizing the matrix over observed user-item preferences. On the other hand, MC methods model sequential behavior by learning a transition graph over items that is used to predict the next action based on the recent actions of a user. In this paper, we present a method bringing both approaches together. Our method is based on personalized transition graphs over underlying Markov chains. That means for each user an own transition matrix is learned - thus in total the method uses a transition cube. As the observations for estimating the transitions are usually very limited, our method factorizes the transition cube with a pairwise interaction model which is a special case of the Tucker Decomposition. We show that our factorized personalized MC (FPMC) model subsumes both a common Markov chain and the normal matrix factorization model. For learning the model parameters, we introduce an adaption of the Bayesian Personalized Ranking (BPR) framework for sequential basket data. Empirically, we show that our FPMC model outperforms both the common matrix factorization and the unpersonalized MC model both learned with and without factorization.",
"Probabilistic topic models are widely used in different contexts to uncover the hidden structure in large text corpora. One of the main (and perhaps strong) assumption of these models is that generative process follows a bag-of-words assumption, i.e. each token is independent from the previous one. We extend the popular Latent Dirichlet Allocation model by exploiting three different conditional Markovian assumptions: (i) the token generation depends on the current topic and on the previous token; (ii) the topic associated with each observation depends on topic associated with the previous one; (iii) the token generation depends on the current and previous topic. For each of these modeling assumptions we present a Gibbs Sampling procedure for parameter estimation. Experimental evaluation over real-word data shows the performance advantages, in terms of recall and precision, of the sequence-modeling approaches.",
"Predicting personalized sequential behavior is a key task for recommender systems. In order to predict user actions such as the next product to purchase, movie to watch, or place to visit, it is essential to take into account both long-term user preferences and sequential patterns (i.e., short-term dynamics). Matrix Factorization and Markov Chain methods have emerged as two separate but powerful paradigms for modeling the two respectively. Combining these ideas has led to unified methods that accommodate long- and short-term dynamics simultaneously by modeling pairwise user-item and item-item interactions. In spite of the success of such methods for tackling dense data, they are challenged by sparsity issues, which are prevalent in real-world datasets. In recent years, similarity-based methods have been proposed for (sequentially-unaware) item recommendation with promising results on sparse datasets. In this paper, we propose to fuse such methods with Markov Chains to make personalized sequential recommendations. We evaluate our method, Fossil, on a variety of large, real-world datasets. We show quantitatively that Fossil outperforms alternative algorithms, especially on sparse datasets, and qualitatively that it captures personalized dynamics and is able to make meaningful recommendations.",
""
]
}
|
1811.09975
|
2949691889
|
Variational autoencoders were proven successful in domains such as computer vision and speech processing. Their adoption for modeling user preferences is still unexplored, although recently it is starting to gain attention in the current literature. In this work, we propose a model which extends variational autoencoders by exploiting the rich information present in the past preference history. We introduce a recurrent version of the VAE, where instead of passing a subset of the whole history regardless of temporal dependencies, we rather pass the consumption sequence subset through a recurrent neural network. At each time-step of the RNN, the sequence is fed through a series of fully-connected layers, the output of which models the probability distribution of the most likely future preferences. We show that handling temporal information is crucial for improving the accuracy of the VAE: In fact, our model beats the current state-of-the-art by valuable margins because of its ability to capture temporal dependencies among the user-consumption sequence using the recurrent encoder still keeping the fundamentals of variational autoencoders intact.
|
Recently, a revamped interest for sequence-based recommendation has taken place, motivated by both the success of recurrent neural networks @cite_0 @cite_32 in domains such as language modeling, and the need to focus on recommendations @cite_41 @cite_51 , i.e. recommendation that do not rely on a user model and instead can cope with single anonymous preference sessions.
|
{
"cite_N": [
"@cite_0",
"@cite_41",
"@cite_51",
"@cite_32"
],
"mid": [
"2950635152",
"2262817822",
"",
"1689711448"
],
"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.",
"We apply recurrent neural networks (RNN) on a new domain, namely recommender systems. Real-life recommender systems often face the problem of having to base recommendations only on short session-based data (e.g. a small sportsware website) instead of long user histories (as in the case of Netflix). In this situation the frequently praised matrix factorization approaches are not accurate. This problem is usually overcome in practice by resorting to item-to-item recommendations, i.e. recommending similar items. We argue that by modeling the whole session, more accurate recommendations can be provided. We therefore propose an RNN-based approach for session-based recommendations. Our approach also considers practical aspects of the task and introduces several modifications to classic RNNs such as a ranking loss function that make it more viable for this specific problem. Experimental results on two data-sets show marked improvements over widely used approaches.",
"",
"Several variants of the long short-term memory (LSTM) architecture for recurrent neural networks have been proposed since its inception in 1995. In recent years, these networks have become the state-of-the-art models for a variety of machine learning problems. This has led to a renewed interest in understanding the role and utility of various computational components of typical LSTM variants. In this paper, we present the first large-scale analysis of eight LSTM variants on three representative tasks: speech recognition, handwriting recognition, and polyphonic music modeling. The hyperparameters of all LSTM variants for each task were optimized separately using random search, and their importance was assessed using the powerful functional ANalysis Of VAriance framework. In total, we summarize the results of 5400 experimental runs ( @math years of CPU time), which makes our study the largest of its kind on LSTM networks. Our results show that none of the variants can improve upon the standard LSTM architecture significantly, and demonstrate the forget gate and the output activation function to be its most critical components. We further observe that the studied hyperparameters are virtually independent and derive guidelines for their efficient adjustment."
]
}
|
1811.09975
|
2949691889
|
Variational autoencoders were proven successful in domains such as computer vision and speech processing. Their adoption for modeling user preferences is still unexplored, although recently it is starting to gain attention in the current literature. In this work, we propose a model which extends variational autoencoders by exploiting the rich information present in the past preference history. We introduce a recurrent version of the VAE, where instead of passing a subset of the whole history regardless of temporal dependencies, we rather pass the consumption sequence subset through a recurrent neural network. At each time-step of the RNN, the sequence is fed through a series of fully-connected layers, the output of which models the probability distribution of the most likely future preferences. We show that handling temporal information is crucial for improving the accuracy of the VAE: In fact, our model beats the current state-of-the-art by valuable margins because of its ability to capture temporal dependencies among the user-consumption sequence using the recurrent encoder still keeping the fundamentals of variational autoencoders intact.
|
@cite_41 proposes a recurrent neural network model based on to predict the next item in a user session, based on the history seen so far. Since its introduction, this model has witnessed several evolutions, and similar architectures were proposed in @cite_16 @cite_40 @cite_21 @cite_14 @cite_48 @cite_30 @cite_2 . Recurrent networks were also exploited to strengthen matrix factorization, by producing history-aware embeddings of users and items. The approach proposed in @cite_22 combines two recurrent networks, whose output at any time-step, relative to a specific user and item, can be hence exploited to predict the current preference.
|
{
"cite_N": [
"@cite_30",
"@cite_14",
"@cite_22",
"@cite_41",
"@cite_48",
"@cite_21",
"@cite_40",
"@cite_2",
"@cite_16"
],
"mid": [
"2746011824",
"",
"2583674722",
"2262817822",
"2953316038",
"2513020047",
"",
"2625746539",
"2726499916"
],
"abstract": [
"Deep learning methods have led to substantial progress in various application fields of AI, and in recent years a number of proposals were made to improve recommender systems with artificial neural networks. For the problem of making session-based recommendations, i.e., for recommending the next item in an anonymous session, recently investigated the application of recurrent neural networks with Gated Recurrent Units (GRU4REC). Assessing the true effectiveness of such novel approaches based only on what is reported in the literature is however difficult when no standard evaluation protocols are applied and when the strength of the baselines used in the performance comparison is not clear. In this work we show based on a comprehensive empirical evaluation that a heuristics-based nearest neighbor (kNN) scheme for sessions outperforms GRU4REC in the large majority of the tested configurations and datasets. Neighborhood sampling and efficient in-memory data structures ensure the scalability of the kNN method. The best results in the end were often achieved when we combine the kNN approach with GRU4REC, which shows that RNNs can leverage sequential signals in the data that cannot be detected by the co-occurrence-based kNN method.",
"",
"Recommender systems traditionally assume that user profiles and movie attributes are static. Temporal dynamics are purely reactive, that is, they are inferred after they are observed, e.g. after a user's taste has changed or based on hand-engineered temporal bias corrections for movies. We propose Recurrent Recommender Networks (RRN) that are able to predict future behavioral trajectories. This is achieved by endowing both users and movies with a Long Short-Term Memory (LSTM) autoregressive model that captures dynamics, in addition to a more traditional low-rank factorization. On multiple real-world datasets, our model offers excellent prediction accuracy and it is very compact, since we need not learn latent state but rather just the state transition function.",
"We apply recurrent neural networks (RNN) on a new domain, namely recommender systems. Real-life recommender systems often face the problem of having to base recommendations only on short session-based data (e.g. a small sportsware website) instead of long user histories (as in the case of Netflix). In this situation the frequently praised matrix factorization approaches are not accurate. This problem is usually overcome in practice by resorting to item-to-item recommendations, i.e. recommending similar items. We argue that by modeling the whole session, more accurate recommendations can be provided. We therefore propose an RNN-based approach for session-based recommendations. Our approach also considers practical aspects of the task and introduces several modifications to classic RNNs such as a ranking loss function that make it more viable for this specific problem. Experimental results on two data-sets show marked improvements over widely used approaches.",
"Recurrent neural networks (RNNs) were recently proposed for the session-based recommendation task. The models showed promising improvements over traditional recommendation approaches. In this work, we further study RNN-based models for session-based recommendations. We propose the application of two techniques to improve model performance, namely, data augmentation, and a method to account for shifts in the input data distribution. We also empirically study the use of generalised distillation, and a novel alternative model that directly predicts item embeddings. Experiments on the RecSys Challenge 2015 dataset demonstrate relative improvements of 12.8 and 14.8 over previously reported results on the Recall@20 and Mean Reciprocal Rank@20 metrics respectively.",
"Preparing recommendations for unknown users or such that correctly respond to the short-term needs of a particular user is one of the fundamental problems for e-commerce. Most of the common Recommender Systems assume that user identification must be explicit. In this paper a Session-Aware Recommender System approach is presented where no straightforward user information is required. The recommendation process is based only on user activity within a single session, defined as a sequence of events. This information is incorporated in the recommendation process by explicit context modeling with factorization methods and a novel approach with Recurrent Neural Network (RNN). Compared to the session modeling approach, RNN directly models the dependency of user observed sequential behavior throughout its recurrent structure. The evaluation discusses the results based on sessions from real-life system with ephemeral items (identified only by the set of their attributes) for the task of top-n best recommendations.",
"",
"Session-based recommendations are highly relevant in many modern on-line services (e.g. e-commerce, video streaming) and recommendation settings. Recently, Recurrent Neural Networks have been shown to perform very well in session-based settings. While in many session-based recommendation domains user identifiers are hard to come by, there are also domains in which user profiles are readily available. We propose a seamless way to personalize RNN models with cross-session information transfer and devise a Hierarchical RNN model that relays end evolves latent hidden states of the RNNs across user sessions. Results on two industry datasets show large improvements over the session-only RNNs.",
"Recurrent neural networks have recently been successfully applied to the session-based recommendation problem, and is part of a growing interest for collaborative filtering based on sequence prediction. This new approach to recommendations reveals an aspect that was previously overlooked: the difference between short-term and long-term recommendations. In this work we characterize the full short-term long-term profile of many collaborative filtering methods, and we show how recurrent neural networks can be steered towards better short or long-term predictions. We also show that RNNs are not only adapted to session-based collaborative filtering, but are perfectly suited for collaborative filtering on dense datasets where it outperforms traditional item recommendation algorithms."
]
}
|
1811.09975
|
2949691889
|
Variational autoencoders were proven successful in domains such as computer vision and speech processing. Their adoption for modeling user preferences is still unexplored, although recently it is starting to gain attention in the current literature. In this work, we propose a model which extends variational autoencoders by exploiting the rich information present in the past preference history. We introduce a recurrent version of the VAE, where instead of passing a subset of the whole history regardless of temporal dependencies, we rather pass the consumption sequence subset through a recurrent neural network. At each time-step of the RNN, the sequence is fed through a series of fully-connected layers, the output of which models the probability distribution of the most likely future preferences. We show that handling temporal information is crucial for improving the accuracy of the VAE: In fact, our model beats the current state-of-the-art by valuable margins because of its ability to capture temporal dependencies among the user-consumption sequence using the recurrent encoder still keeping the fundamentals of variational autoencoders intact.
|
Finally, the CASER () model @cite_27 proposes an approach that departs from RNN modeling and instead exploits a , by transforming a sequence into a matrix built from the concatenation of the embeddings of the items appearing in the sequence. The matrix can hence feed convolutional layers that can extract relevant useful features for predicting the next items.
|
{
"cite_N": [
"@cite_27"
],
"mid": [
"2783272285"
],
"abstract": [
"Top-N sequential recommendation models each user as a sequence of items interacted in the past and aims to predict top-N ranked items that a user will likely interact in a »near future». The order of interaction implies that sequential patterns play an important role where more recent items in a sequence have a larger impact on the next item. In this paper, we propose a Convolutional Sequence Embedding Recommendation Model »Caser» as a solution to address this requirement. The idea is to embed a sequence of recent items into an »image» in the time and latent spaces and learn sequential patterns as local features of the image using convolutional filters. This approach provides a unified and flexible network structure for capturing both general preferences and sequential patterns. The experiments on public data sets demonstrated that Caser consistently outperforms state-of-the-art sequential recommendation methods on a variety of common evaluation metrics."
]
}
|
1811.10302
|
2900551066
|
For visual tracking, most of the traditional correlation filters (CF) based methods suffer from the bottleneck of feature redundancy and lack of motion information. In this paper, we design a novel tracking framework, called multi-hierarchical independent correlation filters (MHIT). The framework consists of motion estimation module, hierarchical features selection, independent CF online learning, and adaptive multi-branch CF fusion. Specifically, the motion estimation module is introduced to capture motion information, which effectively alleviates the object partial occlusion in the temporal video. The multi-hierarchical deep features of CNN representing different semantic information can be fully excavated to track multi-scale objects. To better overcome the deep feature redundancy, each hierarchical features are independently fed into a single branch to implement the online learning of parameters. Finally, an adaptive weight scheme is integrated into the framework to fuse these independent multi-branch CFs for the better and more robust visual object tracking. Extensive experiments on OTB and VOT datasets show that the proposed MHIT tracker can significantly improve the tracking performance. Especially, it obtains a 20.1 relative performance gain compared to the top trackers on the VOT2017 challenge, and also achieves new state-of-the-art performance on the VOT2018 challenge.
|
Based on the CNN methods, @cite_4 exploit the CNN end-to-end training approach to turn the tracking problem into a classification problem. MDNet @cite_28 further combines offline multi-domain training and online updates classifiers for identifying specific targets. Following the end-to-end ideas, some works further use a Siamese matching structure to learn a similarity measure, which regards DCF as part of the networks. SiamFC @cite_38 trains offline with the ILSVRC @cite_32 dataset and does not update the parameters online. DCFNet @cite_33 presents an end-to-end network architecture to learn the convolutional features and performs the correlation tracking process simultaneously. SiamRPN @cite_21 introduces feature extraction and region proposal subnetwork including the classification branch and regression branch. DaSiamRPN @cite_13 proposes a framework on the basis of SiamRPN @cite_21 to learn distractor-aware features and explicitly suppress distractors during the inference of online tracking. SiamVGG @cite_12 replaces the base network AlexNet @cite_3 with VGG @cite_16 on the basis of SiamFC @cite_38 to improve tracking performance. This type of methods typically takes the groundtruth of the first frame as a template or employs a simple moving average strategy to update the template.
|
{
"cite_N": [
"@cite_38",
"@cite_4",
"@cite_33",
"@cite_28",
"@cite_21",
"@cite_32",
"@cite_3",
"@cite_16",
"@cite_13",
"@cite_12"
],
"mid": [
"",
"1554825167",
"2605381261",
"1857884451",
"2799058067",
"2117539524",
"",
"1686810756",
"2886910176",
"2913466142"
],
"abstract": [
"",
"Deep neural networks, albeit their great success on feature learning in various computer vision tasks, are usually considered as impractical for online visual tracking, because they require very long training time and a large number of training samples. In this paper, we present an efficient and very robust tracking algorithm using a single convolutional neural network (CNN) for learning effective feature representations of the target object in a purely online manner. Our contributions are multifold. First, we introduce a novel truncated structural loss function that maintains as many training samples as possible and reduces the risk of tracking error accumulation. Second, we enhance the ordinary stochastic gradient descent approach in CNN training with a robust sample selection mechanism. The sampling mechanism randomly generates positive and negative samples from different temporal distributions, which are generated by taking the temporal relations and label noise into account. Finally, a lazy yet effective updating scheme is designed for CNN training. Equipped with this novel updating algorithm, the CNN model is robust to some long-existing difficulties in visual tracking, such as occlusion or incorrect detections, without loss of the effective adaption for significant appearance changes. In the experiment, our CNN tracker outperforms all compared state-of-the-art methods on two recently proposed benchmarks, which in total involve over 60 video sequences. The remarkable performance improvement over the existing trackers illustrates the superiority of the feature representations, which are learned purely online via the proposed deep learning framework.",
"Discriminant Correlation Filters (DCF) based methods now become a kind of dominant approach to online object tracking. The features used in these methods, however, are either based on hand-crafted features like HoGs, or convolutional features trained independently from other tasks like image classification. In this work, we present an end-to-end lightweight network architecture, namely DCFNet, to learn the convolutional features and perform the correlation tracking process simultaneously. Specifically, we treat DCF as a special correlation filter layer added in a Siamese network, and carefully derive the backpropagation through it by defining the network output as the probability heatmap of object location. Since the derivation is still carried out in Fourier frequency domain, the efficiency property of DCF is preserved. This enables our tracker to run at more than 60 FPS during test time, while achieving a significant accuracy gain compared with KCF using HoGs. Extensive evaluations on OTB-2013, OTB-2015, and VOT2015 benchmarks demonstrate that the proposed DCFNet tracker is competitive with several state-of-the-art trackers, while being more compact and much faster.",
"We propose a novel visual tracking algorithm based on the representations from a discriminatively trained Convolutional Neural Network (CNN). Our algorithm pretrains a CNN using a large set of videos with tracking groundtruths to obtain a generic target representation. Our network is composed of shared layers and multiple branches of domain-specific layers, where domains correspond to individual training sequences and each branch is responsible for binary classification to identify target in each domain. We train each domain in the network iteratively to obtain generic target representations in the shared layers. When tracking a target in a new sequence, we construct a new network by combining the shared layers in the pretrained CNN with a new binary classification layer, which is updated online. Online tracking is performed by evaluating the candidate windows randomly sampled around the previous target state. The proposed algorithm illustrates outstanding performance in existing tracking benchmarks.",
"Visual object tracking has been a fundamental topic in recent years and many deep learning based trackers have achieved state-of-the-art performance on multiple benchmarks. However, most of these trackers can hardly get top performance with real-time speed. In this paper, we propose the Siamese region proposal network (Siamese-RPN) which is end-to-end trained off-line with large-scale image pairs. Specifically, it consists of Siamese subnetwork for feature extraction and region proposal subnetwork including the classification branch and regression branch. In the inference phase, the proposed framework is formulated as a local one-shot detection task. We can pre-compute the template branch of the Siamese subnetwork and formulate the correlation layers as trivial convolution layers to perform online tracking. Benefit from the proposal refinement, traditional multi-scale test and online fine-tuning can be discarded. The Siamese-RPN runs at 160 FPS while achieving leading performance in VOT2015, VOT2016 and VOT2017 real-time challenges.",
"The ImageNet Large Scale Visual Recognition Challenge is a benchmark in object category classification and detection on hundreds of object categories and millions of images. The challenge has been run annually from 2010 to present, attracting participation from more than fifty institutions. This paper describes the creation of this benchmark dataset and the advances in object recognition that have been possible as a result. We discuss the challenges of collecting large-scale ground truth annotation, highlight key breakthroughs in categorical object recognition, provide a detailed analysis of the current state of the field of large-scale image classification and object detection, and compare the state-of-the-art computer vision accuracy with human accuracy. We conclude with lessons learned in the 5 years of the challenge, and propose future directions and improvements.",
"",
"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.",
"Recently, Siamese networks have drawn great attention in visual tracking community because of their balanced accuracy and speed. However, features used in most Siamese tracking approaches can only discriminate foreground from the non-semantic backgrounds. The semantic backgrounds are always considered as distractors, which hinders the robustness of Siamese trackers. In this paper, we focus on learning distractor-aware Siamese networks for accurate and long-term tracking. To this end, features used in traditional Siamese trackers are analyzed at first. We observe that the imbalanced distribution of training data makes the learned features less discriminative. During the off-line training phase, an effective sampling strategy is introduced to control this distribution and make the model focus on the semantic distractors. During inference, a novel distractor-aware module is designed to perform incremental learning, which can effectively transfer the general embedding to the current video domain. In addition, we extend the proposed approach for long-term tracking by introducing a simple yet effective local-to-global search region strategy. Extensive experiments on benchmarks show that our approach significantly outperforms the state-of-the-arts, yielding 9.6 relative gain in VOT2016 dataset and 35.9 relative gain in UAV20L dataset. The proposed tracker can perform at 160 FPS on short-term benchmarks and 110 FPS on long-term benchmarks.",
"The Visual Object Tracking challenge VOT2018 is the sixth annual tracker benchmarking activity organized by the VOT initiative. Results of over eighty trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in the recent years. The evaluation included the standard VOT and other popular methodologies for short-term tracking analysis and a “real-time” experiment simulating a situation where a tracker processes images as if provided by a continuously running sensor. A long-term tracking subchallenge has been introduced to the set of standard VOT sub-challenges. The new subchallenge focuses on long-term tracking properties, namely coping with target disappearance and reappearance. A new dataset has been compiled and a performance evaluation methodology that focuses on long-term tracking capabilities has been adopted. The VOT toolkit has been updated to support both standard short-term and the new long-term tracking subchallenges. Performance of the tested trackers typically by far exceeds standard baselines. The source code for most of the trackers is publicly available from the VOT page. The dataset, the evaluation kit and the results are publicly available at the challenge website (http: votchallenge.net)."
]
}
|
1811.10238
|
2901667120
|
Many dialogue management frameworks allow the system designer to directly define belief rules to implement an efficient dialog policy. Because these rules are directly defined, the components are said to be hand-crafted. As dialogues become more complex, the number of states, transitions, and policy decisions becomes very large. To facilitate the dialog policy design process, we propose an approach to automatically learn belief rules using a supervised machine learning approach. We validate our ideas in Student-Advisor conversation domain, where we extract latent beliefs like student is curious, confused and neutral, etc. Further, we also perform epistemic reasoning that helps to tailor the dialog according to student's emotional state and hence improve the overall effectiveness of the dialog system. Our latent belief identification approach shows an accuracy of 87 and this results in efficient and meaningful dialog management.
|
Deep learning based dialog systems @cite_7 use memory networks to learn the underlying dialog structure and carry out goal-oriented dialog. However, they do not factor in beliefs or trigger epistemic rules in modifying the conversation given the evolving context. @cite_1 Williams et.al, describe the dialog state tracking challenge and mention how the task of correctly inferring the state of the conversation - such as the user's goal - given all of the dialog history up to that turn'' is important. It is in this overall context, we propose that it is important to evaluate the probable beliefs held by the human and tailor the dialog system suitably to be consistent with the beliefs in order to hold a relevant conversation.
|
{
"cite_N": [
"@cite_1",
"@cite_7"
],
"mid": [
"2468710617",
"2409591106"
],
"abstract": [
"In a spoken dialog system, dialog state tracking refers to the task of correctly inferring the state of the conversation -- such as the user's goal -- given all of the dialog history up to that turn. Dialog state tracking is crucial to the success of a dialog system, yet until recently there were no common resources, hampering progress. The Dialog State Tracking Challenge series of 3 tasks introduced the first shared testbed and evaluation metrics for dialog state tracking, and has underpinned three key advances in dialog state tracking: the move from generative to discriminative models; the adoption of discriminative sequential techniques; and the incorporation of the speech recognition results directly into the dialog state tracker. This paper reviews this research area, covering both the challenge tasks themselves and summarizing the work they have enabled.",
"Directly reading documents and being able to answer questions from them is an unsolved challenge. To avoid its inherent difficulty, question answering (QA) has been directed towards using Knowledge Bases (KBs) instead, which has proven effective. Unfortunately KBs often suffer from being too restrictive, as the schema cannot support certain types of answers, and too sparse, e.g. Wikipedia contains much more information than Freebase. In this work we introduce a new method, Key-Value Memory Networks, that makes reading documents more viable by utilizing different encodings in the addressing and output stages of the memory read operation. To compare using KBs, information extraction or Wikipedia documents directly in a single framework we construct an analysis tool, WikiMovies, a QA dataset that contains raw text alongside a preprocessed KB, in the domain of movies. Our method reduces the gap between all three settings. It also achieves state-of-the-art results on the existing WikiQA benchmark."
]
}
|
1811.10238
|
2901667120
|
Many dialogue management frameworks allow the system designer to directly define belief rules to implement an efficient dialog policy. Because these rules are directly defined, the components are said to be hand-crafted. As dialogues become more complex, the number of states, transitions, and policy decisions becomes very large. To facilitate the dialog policy design process, we propose an approach to automatically learn belief rules using a supervised machine learning approach. We validate our ideas in Student-Advisor conversation domain, where we extract latent beliefs like student is curious, confused and neutral, etc. Further, we also perform epistemic reasoning that helps to tailor the dialog according to student's emotional state and hence improve the overall effectiveness of the dialog system. Our latent belief identification approach shows an accuracy of 87 and this results in efficient and meaningful dialog management.
|
Although a number of attempts have been made to build dialog systems @cite_9 , @cite_3 , @cite_1 , the use of epistemic rules in driving the dialog in a consistent way with the beliefs has not yet been tackled. Various approaches to dialog management have been proposed and these can be broadly classified into finite-state methods, probabilistic methods and deep learned methods.
|
{
"cite_N": [
"@cite_9",
"@cite_1",
"@cite_3"
],
"mid": [
"2250297846",
"2468710617",
""
],
"abstract": [
"Recently discriminative methods for tracking the state of a spoken dialog have been shown to outperform traditional generative models. This paper presents a new wordbased tracking method which maps directly from the speech recognition results to the dialog state without using an explicit semantic decoder. The method is based on a recurrent neural network structure which is capable of generalising to unseen dialog state hypotheses, and which requires very little feature engineering. The method is evaluated on the second Dialog State Tracking Challenge (DSTC2) corpus and the results demonstrate consistently high performance across all of the metrics.",
"In a spoken dialog system, dialog state tracking refers to the task of correctly inferring the state of the conversation -- such as the user's goal -- given all of the dialog history up to that turn. Dialog state tracking is crucial to the success of a dialog system, yet until recently there were no common resources, hampering progress. The Dialog State Tracking Challenge series of 3 tasks introduced the first shared testbed and evaluation metrics for dialog state tracking, and has underpinned three key advances in dialog state tracking: the move from generative to discriminative models; the adoption of discriminative sequential techniques; and the incorporation of the speech recognition results directly into the dialog state tracker. This paper reviews this research area, covering both the challenge tasks themselves and summarizing the work they have enabled.",
""
]
}
|
1811.09955
|
2900903413
|
Online learning with limited information feedback (bandit) tries to solve the problem where an online learner receives partial feedback information from the environment in the course of learning. Under this setting, [8] extended Zinkevich's classical Online Gradient Descent (OGD) algorithm [29] by proposing the Online Gradient Descent with Expected Gradient (OGDEG) algorithm. Specifically, it uses a simple trick to approximate the gradient of the loss function @math by evaluating it at a single point and bounds the expected regret as @math [8], where the number of rounds is @math . Meanwhile, past research efforts have shown that compared with the first-order algorithms, second-order online learning algorithms such as Online Newton Step (ONS) [11] can significantly accelerate the convergence rate of traditional online learning algorithms. Motivated by this, this paper aims to exploit the second-order information to speed up the convergence of the OGDEG algorithm. In particular, we extend the ONS algorithm with the trick of expected gradient and develop a novel second-order online learning algorithm, i.e., Online Newton Step with Expected Gradient (ONSEG). Theoretically, we show that the proposed ONSEG algorithm significantly reduces the expected regret of OGDEG algorithm from @math to @math in the bandit feedback scenario. Empirically, we further demonstrate the advantages of the proposed algorithm on multiple real-world datasets.
|
In this section, we briefly review some related work on bandit convex optimization algorithms which are closely related to our ONSEG. The study of bandit convex optimization was pioneered by @cite_31 and @cite_14 . One key insight into the bandit convex optimization problem @cite_14 , is that the subgradient of a smoothed version of the loss function can be estimated by sampling and rescaling around the points the algorithm originally intended to operate on. Flaxman proved that a gradient descent-type strategy with a one-point estimate of the gradient achieve expected regret bound as @math for convex bounded loss. For the setting of Lipschitz-continuous convex loss, a bound of @math were obtained by @cite_31 and @cite_14 . To improve the regret bound of bandit online learning, numerous learning algorithms have been proposed. Among them, @cite_33 proposed the Geometric Hedge algorithm that achieves an optimal regret bound of @math for linear loss functions. Inspired by the interior point methods, @cite_21 devised an algorithm that attains the same nearly-optimal regret bound for bandit linear optimization.
|
{
"cite_N": [
"@cite_31",
"@cite_14",
"@cite_21",
"@cite_33"
],
"mid": [
"2097487180",
"2952840318",
"",
"2120745256"
],
"abstract": [
"In the multi-armed bandit problem, an online algorithm must choose from a set of strategies in a sequence of n trials so as to minimize the total cost of the chosen strategies. While nearly tight upper and lower bounds are known in the case when the strategy set is finite, much less is known when there is an infinite strategy set. Here we consider the case when the set of strategies is a subset of ℝd, and the cost functions are continuous. In the d = 1 case, we improve on the best-known upper and lower bounds, closing the gap to a sublogarithmic factor. We also consider the case where d > 1 and the cost functions are convex, adapting a recent online convex optimization algorithm of Zinkevich to the sparser feedback model of the multi-armed bandit problem.",
"We consider a the general online convex optimization framework introduced by Zinkevich. In this setting, there is a sequence of convex functions. Each period, we must choose a signle point (from some feasible set) and pay a cost equal to the value of the next function on our chosen point. Zinkevich shows that, if the each function is revealed after the choice is made, then one can achieve vanishingly small regret relative the best single decision chosen in hindsight. We extend this to the bandit setting where we do not find out the entire functions but rather just their value at our chosen point. We show how to get vanishingly small regret in this setting. Our approach uses a simple approximation of the gradient that is computed from evaluating a function at a single (random) point. We show that this estimate is sufficient to mimic Zinkevich's gradient descent online analysis, with access to the gradient (only being able to evaluate the function at a single point).",
"",
"In the online linear optimization problem, a learner must choose, in each round, a decision from a set D ⊂ ℝn in order to minimize an (unknown and changing) linear cost function. We present sharp rates of convergence (with respect to additive regret) for both the full information setting (where the cost function is revealed at the end of each round) and the bandit setting (where only the scalar cost incurred is revealed). In particular, this paper is concerned with the price of bandit information, by which we mean the ratio of the best achievable regret in the bandit setting to that in the full-information setting. For the full information case, the upper bound on the regret is O*( √nT), where n is the ambient dimension and T is the time horizon. For the bandit case, we present an algorithm which achieves O*(n3 2 √T) regret — all previous (nontrivial) bounds here were O(poly(n)T2 3) or worse. It is striking that the convergence rate for the bandit setting is only a factor of n worse than in the full information case — in stark contrast to the K-arm bandit setting, where the gap in the dependence on K is exponential (√TK vs. √T log K). We also present lower bounds showing that this gap is at least √n, which we conjecture to be the correct order. The bandit algorithm we present can be implemented efficiently in special cases of particular interest, such as path planning and Markov Decision Problems."
]
}
|
1811.09955
|
2900903413
|
Online learning with limited information feedback (bandit) tries to solve the problem where an online learner receives partial feedback information from the environment in the course of learning. Under this setting, [8] extended Zinkevich's classical Online Gradient Descent (OGD) algorithm [29] by proposing the Online Gradient Descent with Expected Gradient (OGDEG) algorithm. Specifically, it uses a simple trick to approximate the gradient of the loss function @math by evaluating it at a single point and bounds the expected regret as @math [8], where the number of rounds is @math . Meanwhile, past research efforts have shown that compared with the first-order algorithms, second-order online learning algorithms such as Online Newton Step (ONS) [11] can significantly accelerate the convergence rate of traditional online learning algorithms. Motivated by this, this paper aims to exploit the second-order information to speed up the convergence of the OGDEG algorithm. In particular, we extend the ONS algorithm with the trick of expected gradient and develop a novel second-order online learning algorithm, i.e., Online Newton Step with Expected Gradient (ONSEG). Theoretically, we show that the proposed ONSEG algorithm significantly reduces the expected regret of OGDEG algorithm from @math to @math in the bandit feedback scenario. Empirically, we further demonstrate the advantages of the proposed algorithm on multiple real-world datasets.
|
For some specifical classes of nonlinear convex losses, several methods have been proposed @cite_11 @cite_23 @cite_0 @cite_1 @cite_13 . Under the assumption of strongly convex loss, @cite_7 attained an upper bound of @math . The follow-up work of Saha @cite_11 showed that for convex and smooth loss functions, one can run FTRL with a self-concordant barrier as regularization and sample around the Dikin ellipsoid to attain regret bound of @math . In a recent paper, Hazan and Levy investigated the bandit convex optimization setting by assuming that the adversary is limited to inflicting strongly convex and smooth loss functions and the player may choose points from a constrained set. In this setting, they devised an algorithm that achieves a regret bound of @math . While a recent paper by @cite_2 shows that the lower bound of regret has to be @math even with the strongly convex and smooth assumptions.
|
{
"cite_N": [
"@cite_7",
"@cite_1",
"@cite_0",
"@cite_23",
"@cite_2",
"@cite_13",
"@cite_11"
],
"mid": [
"153281708",
"",
"2556949899",
"2473549844",
"2963062786",
"2151056989",
"2284345772"
],
"abstract": [
"Bandit convex optimization is a special case of online convex optimization with partial information. In this setting, a player attempts to minimize a sequence of adversarially generated convex loss functions, while only observing the value of each function at a single point. In some cases, the minimax regret of these problems is known to be strictly worse than the minimax regret in the corresponding full information setting. We introduce the multi-point bandit setting, in which the player can query each loss function at multiple points. When the player is allowed to query each function at two points, we prove regret bounds that closely resemble bounds for the full information case. This suggests that knowing the value of each loss function at two points is almost as useful as knowing the value of each function everywhere. When the player is allowed to query each function at d+1 points (d being the dimension of the space), we prove regret bounds that are exactly equivalent to full information bounds for smooth functions.",
"",
"We introduce the general and powerful scheme of predicting information re-use in optimization algorithms. This allows us to devise a computationally efficient algorithm for bandit convex optimization with new state-of-the-art guarantees for both Lipschitz loss functions and loss functions with Lipschitz gradients. This is the first algorithm admitting both a polynomial time complexity and a regret that is polynomial in the dimension of the action space that improves upon the original regret bound for Lipschitz loss functions, achieving a regret of @math . Our algorithm further improves upon the best existing polynomial-in-dimension bound (both computationally and in terms of regret) for loss functions with Lipschitz gradients, achieving a regret of @math .",
"We consider the adversarial convex bandit problem and we build the first @math -time algorithm with @math -regret for this problem. To do so we introduce three new ideas in the derivative-free optimization literature: (i) kernel methods, (ii) a generalization of Bernoulli convolutions, and (iii) a new annealing schedule for exponential weights (with increasing learning rate). The basic version of our algorithm achieves @math -regret, and we show that a simple variant of this algorithm can be run in @math -time per step at the cost of an additional @math factor in the regret. These results improve upon the @math -regret and @math -time result of the first two authors, and the @math -regret and @math -time result of Hazan and Li. Furthermore we conjecture that another variant of the algorithm could achieve @math -regret, and moreover that this regret is unimprovable (the current best lower bound being @math and it is achieved with linear functions). For the simpler situation of zeroth order stochastic convex optimization this corresponds to the conjecture that the optimal query complexity is of order @math .",
"The problem of stochastic convex optimization with bandit feedback (in the learning community) or without knowledge of gradients (in the optimization community) has received much attention in recent years, in the form of algorithms and performance upper bounds. However, much less is known about the inherent complexity of these problems, and there are few lower bounds in the literature, especially for nonlinear functions. In this paper, we investigate the attainable error regret in the bandit and derivative-free settings, as a function of the dimension d and the available number of queries T . We provide a precise characterization of the attainable performance for strongly-convex and smooth functions, which also imply a non-trivial lower bound for more general problems. Moreover, we prove that in both the bandit and derivative-free setting, the required number of queries must scale at least quadratically with the dimension. Finally, we show that on the natural class of quadratic functions, it is possible to obtain a \"O(1=T ) error rate in terms of T , under mild assumptions, even without having access to gradients. To the best of our knowledge, this is the rst such rate in a derivative-free stochastic setting, and holds despite previous",
"Bandit Convex Optimization (BCO) is a fundamental framework for decision making under uncertainty, which generalizes many problems from the realm of online and statistical learning. While the special case of linear cost functions is well understood, a gap on the attainable regret for BCO with nonlinear losses remains an important open question. In this paper we take a step towards understanding the best attainable regret bounds for BCO: we give an efficient and near-optimal regret algorithm for BCO with strongly-convex and smooth loss functions. In contrast to previous works on BCO that use time invariant exploration schemes, our method employs an exploration scheme that shrinks with time.",
"The study of online convex optimization in the bandit setting was initiated by Kleinberg (2004) and (2005). Such a setting models a decision maker that has to make decisions in the face of adversarially chosen convex loss functions. Moreover, the only information the decision maker receives are the losses. The identities of the loss functions themselves are not revealed. In this setting, we reduce the gap between the best known lower and upper bounds for the class of smooth convex functions, i.e. convex functions with a Lipschitz continuous gradient. Building upon existing work on selfconcordant regularizers and one-point gradient estimation, we give the first algorithm whose expected regret is O(T ), ignoring constant and logarithmic factors."
]
}
|
1811.10276
|
2901571171
|
Using variational autoencoders trained on known physics processes, we develop a one-sided threshold test to isolate previously unseen processes as outlier events. Since the autoencoder training does not depend on any specific new physics signature, the proposed procedure doesn't make specific assumptions on the nature of new physics. An event selection based on this algorithm would be complementary to classic LHC searches, typically based on model-dependent hypothesis testing. Such an algorithm would deliver a list of anomalous events, that the experimental collaborations could further scrutinize and even release as a catalog, similarly to what is typically done in other scientific domains. Event topologies repeating in this dataset could inspire new-physics model building and new experimental searches. Running in the trigger system of the LHC experiments, such an application could identify anomalous events that would be otherwise lost, extending the scientific reach of the LHC.
|
Model-independent searches for new physics have been performed at the Tevatron @cite_16 @cite_22 , at HERA @cite_12 , and the LHC @cite_31 @cite_15 . These searches are based on the comparison of a large set of binned distributions to the prediction from Monte Carlo simulation, in search for bins exhibiting a deviation larger than some predefined threshold. While the effectiveness of this strategy in establishing a discovery has been matter of discussion, a recent study by the ATLAS collaboration @cite_15 has rephrased this model-independent search strategy into a tool to identify interesting excesses, on which traditional analysis techniques could be performed on independent datasets (e.g., the data collected after running the model-independent analysis). This change of scope has the advantage of reducing the trial factor (i.e., the so-called look-elsewhere effect @cite_17 @cite_11 ), which washes out the significance of an observed excess.
|
{
"cite_N": [
"@cite_11",
"@cite_22",
"@cite_15",
"@cite_31",
"@cite_16",
"@cite_12",
"@cite_17"
],
"mid": [
"2086157270",
"2324761965",
"2884241072",
"",
"2135757595",
"2131231395",
"1971267516"
],
"abstract": [
"When searching for a new resonance somewhere in a possible mass range, the significance of observing a local excess of events must take into account the probability of observing such an excess anywhere in the range. This is the so called “look elsewhere effect”. The effect can be quantified in terms of a trial factor, which is the ratio between the probability of observing the excess at some fixed mass point, to the probability of observing it anywhere in the range. We propose a simple and fast procedure for estimating the trial factor, based on earlier results by Davies. We show that asymptotically, the trial factor grows linearly with the (fixed mass) significance.",
"We describe a model independent search for physics beyond the standard model in lepton final states. We examine 117 final states using 1.1 fb @math of @math collisions data at @math TeV collected with the D0 detector. We conclude that all observed discrepancies between data and model can be attributed to uncertainties in the standard model background modeling, and hence we do not see any evidence for physics beyond the standard model.",
"This paper describes a strategy for a general search used by the ATLAS Collaboration to find potential indications of new physics. Events are classified according to their final state into many event classes. For each event class an automated search algorithm tests whether the data are compatible with the Monte Carlo simulated expectation in several distributions sensitive to the effects of new physics. The significance of a deviation is quantified using pseudo-experiments. A data selection with a significant deviation defines a signal region for a dedicated follow-up analysis with an improved background expectation. The analysis of the data-derived signal regions on a new dataset allows a statistical interpretation without the large look-elsewhere effect. The sensitivity of the approach is discussed using Standard Model processes and benchmark signals of new physics. As an example, results are shown for 3.2 fb −1 of proton–proton collision data at a centre-of-mass energy of 13 TeV collected with the ATLAS detector at the LHC in 2015, in which more than 700 event classes and more than 105 regions have been analysed. No significant deviations are found and consequently no data-derived signal regions for a follow-up analysis have been defined.",
"",
"Data collected in run II of the Fermilab Tevatron are searched for indications of new electroweak-scale physics. Rather than focusing on particular new physics scenarios, CDF data are analyzed for discrepancies with the standard model prediction. A model-independent approach (VISTA) considers gross features of the data, and is sensitive to new large cross-section physics. Further sensitivity to new physics is provided by two additional algorithms: a Bump Hunter searches invariant mass distributions for \"bumps'' that could indicate resonant production of new particles, and the SLEUTH procedure scans for data excesses at large summed transverse momentum. This combined global search for new physics in 2.0 fb(-1) of p (p) over bar collisions at root s = 1.96 TeV reveals no indication of physics beyond the standard model.",
"A model--independent search for deviations from the Standard Model prediction is performed using the full @math data sample collected by the H1 experiment at HERA. All event topologies involving isolated electrons, photons, muons, neutrinos and jets with transverse momenta above 20 GeV are investigated in a single analysis. Events are assigned to exclusive classes according to their final state. A dedicated algorithm is used to search for deviations from the Standard Model in the distributions of the scalar sum of transverse momenta or the invariant mass of final state particles and to quantify their significance. Variables related to angular distributions and energy sharing between final state particles are also introduced to study the final state topologies. No significant deviation from the Standard Model expectation is observed in the phase space covered by this analysis.",
"Many statistical issues arise in the analysis of Particle Physics experiments. We give a brief introduction to Particle Physics, before describing the techniques used by Particle Physicists for dealing with statistical problems, and also some of the open statistical questions."
]
}
|
1811.10276
|
2901571171
|
Using variational autoencoders trained on known physics processes, we develop a one-sided threshold test to isolate previously unseen processes as outlier events. Since the autoencoder training does not depend on any specific new physics signature, the proposed procedure doesn't make specific assumptions on the nature of new physics. An event selection based on this algorithm would be complementary to classic LHC searches, typically based on model-dependent hypothesis testing. Such an algorithm would deliver a list of anomalous events, that the experimental collaborations could further scrutinize and even release as a catalog, similarly to what is typically done in other scientific domains. Event topologies repeating in this dataset could inspire new-physics model building and new experimental searches. Running in the trigger system of the LHC experiments, such an application could identify anomalous events that would be otherwise lost, extending the scientific reach of the LHC.
|
Our strategy is similar to what is proposed in Ref. @cite_15 , with two substantial differences: (i) we aim to monitor also those events that could be discarded by the online selection, by running the algorithm in the trigger system; (ii) we do so exploiting deep-learning-based anomaly detection techniques.
|
{
"cite_N": [
"@cite_15"
],
"mid": [
"2884241072"
],
"abstract": [
"This paper describes a strategy for a general search used by the ATLAS Collaboration to find potential indications of new physics. Events are classified according to their final state into many event classes. For each event class an automated search algorithm tests whether the data are compatible with the Monte Carlo simulated expectation in several distributions sensitive to the effects of new physics. The significance of a deviation is quantified using pseudo-experiments. A data selection with a significant deviation defines a signal region for a dedicated follow-up analysis with an improved background expectation. The analysis of the data-derived signal regions on a new dataset allows a statistical interpretation without the large look-elsewhere effect. The sensitivity of the approach is discussed using Standard Model processes and benchmark signals of new physics. As an example, results are shown for 3.2 fb −1 of proton–proton collision data at a centre-of-mass energy of 13 TeV collected with the ATLAS detector at the LHC in 2015, in which more than 700 event classes and more than 105 regions have been analysed. No significant deviations are found and consequently no data-derived signal regions for a follow-up analysis have been defined."
]
}
|
1811.10276
|
2901571171
|
Using variational autoencoders trained on known physics processes, we develop a one-sided threshold test to isolate previously unseen processes as outlier events. Since the autoencoder training does not depend on any specific new physics signature, the proposed procedure doesn't make specific assumptions on the nature of new physics. An event selection based on this algorithm would be complementary to classic LHC searches, typically based on model-dependent hypothesis testing. Such an algorithm would deliver a list of anomalous events, that the experimental collaborations could further scrutinize and even release as a catalog, similarly to what is typically done in other scientific domains. Event topologies repeating in this dataset could inspire new-physics model building and new experimental searches. Running in the trigger system of the LHC experiments, such an application could identify anomalous events that would be otherwise lost, extending the scientific reach of the LHC.
|
Recent works @cite_27 @cite_30 @cite_18 @cite_24 have investigated the use of machine-learning techniques to setup new strategies for BSM searches with minimal or no assumption on the specific new-physics scenario under investigation. In this work, we use variational autoencoders based on high-level features as a baseline. Previously, autoencoders have been used in collider physics for detector monitoring @cite_2 @cite_29 and event generation @cite_3 . Autoencoders have also been explored to define a jet tagger that would identify new physics events with anomalous jets @cite_32 @cite_25 , with a strategy similar to what we apply to the full event in this work.
|
{
"cite_N": [
"@cite_30",
"@cite_18",
"@cite_29",
"@cite_32",
"@cite_3",
"@cite_24",
"@cite_27",
"@cite_2",
"@cite_25"
],
"mid": [
"2801566192",
"2884928202",
"2903631304",
"2888983454",
"2899415024",
"2883371446",
"2807595580",
"2885353798",
"2889504592"
],
"abstract": [
"The oldest and most robust technique to search for new particles is to look for bumps' in invariant mass spectra over smoothly falling backgrounds. This is a powerful technique, but only uses one-dimensional information. One can restrict the phase space to enhance a potential signal, but such tagging techniques require a signal hypothesis and training a classifier in simulation and applying it on data. We present a new method for using all of the available information (with machine learning) without using any prior knowledge about potential signals. Given the lack of new physics signals at the Large Hadron Collider (LHC), such model independent approaches are critical for ensuring full coverage to fully exploit the rich datasets from the LHC experiments. In addition to illustrating how the new method works in simple test cases, we demonstrate the power of the extended bump hunt on a realistic all-hadronic resonance search in a channel that would not be covered with existing techniques.",
"We propose a new scientific application of unsupervised learning techniques to boost our ability to search for new phenomena in data, by detecting discrepancies between two datasets. These could be, for example, a simulated standard-model background, and an observed dataset containing a potential hidden signal of New Physics. We build a statistical test upon a test statistic which measures deviations between two samples, using a Nearest Neighbors approach to estimate the local ratio of the density of points. The test is model-independent and non-parametric, requiring no knowledge of the shape of the underlying distributions, and it does not bin the data, thus retaining full information from the multidimensional feature space. As a proof-of-concept, we apply our method to synthetic Gaussian data, and to a simulated dark matter signal at the Large Hadron Collider. Even in the case where the background can not be simulated accurately enough to claim discovery, the technique is a powerful tool to identify regions of interest for further study.",
"",
"Autoencoder networks, trained only on QCD jets, can be used to search for anomalies in jet-substructure. We show how, based either on images or on 4-vectors, they identify jets from decays of arbitrary heavy resonances. To control the backgrounds and the underlying systematics we can de-correlate the jet mass using an adversarial network. Such an adversarial autoencoder allows for a general and at the same time easily controllable search for new physics. Ideally, it can be trained and applied to data in the same phase space region, allowing us to efficiently search for new physics using un-supervised learning.",
"Detectors of High Energy Physics experiments, such as the ATLAS dectector [1] at the Large Hadron Collider [2], serve as cameras that take pictures of the particles produced in the collision events. One of the key detector technologies used for measuring the energy of particles are calorimeters. Particles will lose their energy in a cascade (called a shower) of electromagnetic and hadronic interactions with a dense absorbing material. The number of the particles produced in this showering process is subsequently measured across the sampling layers of the calorimeter. The deposition of energy in the calorimeter due to a developing shower is a stochastic process that can not be described from first principles and rather relies on a precise simulation of the detector response. It requires the modeling of particles interactions with matter at the microscopic level as implemented using the Geant4 toolkit [3]. This simulation process is inherently slow and thus presents a bottleneck in the ATLAS simulation pipeline [4]. The current work addresses this limitation. To meet the growing analysis demands, ATLAS already relies strongly on fast calorimeter simulation techniques based on thousands of individual parametrizations of the calorimeter response [5]. The algorithms currently employed for physics analyses by the ATLAS collaboration achieve a significant speedup over the full simulation of the detector response at the cost of accuracy. Current developments [6] [7] aim at improving the modeling of taus, jet-substructure-based boosted objects or wrongly identified objects in the calorimeter and will benefit from an improved detector description following data taking and a more detailed forward calorimeter geometry. Deep Learning techniques have been improving state of the art results in various science areas such as: astrophysics [8], cosmology [9] and medical imaging [10]. These techniques are able to describe complex data structures and scale well with highdimensionality problems. Generative models are powerful deep learning algorithms to map complex distributions into a lower dimensional space, to generate samples of higher dimensionality and to approximate the underlying probability densities. Among the most promising approaches are Variational Auto-Encoders [11] [12] and Generative Adversarial Networks [13]. In this context, the talk presents the first application of such models to the fast simulation of the calorimeter response in the ATLAS detector. This work [14] demonstrates the feasibility of using such algorithms for large scale high energy physics experiments in the future, and opens the possibility to complement current techniques.",
"Novelty detection is the machine learning task to recognize data, which belong to an unknown pattern. Complementary to supervised learning, it allows to analyze data model-independently. We demonstrate the potential role of novelty detection in collider physics, using autoencoder-based deep neural network. Explicitly, we develop a set of density-based novelty evaluators, which are sensitive to the clustering of unknown-pattern testing data or new-physics signal events, for the design of detection algorithms. We also explore the influence of the known-pattern data fluctuations, arising from non-signal regions, on detection sensitivity. Strategies to address it are proposed. The algorithms are applied to detecting fermionic di-top partner and resonant di-top productions at LHC, and exotic Higgs decays of two specific modes at a @math future collider. With parton-level analysis, we conclude that potentially the new-physics benchmarks can be recognized with high efficiency.",
"We propose using neural networks to detect data departures from a given reference model, with no prior bias on the nature of the new physics responsible for the discrepancy. The virtues of neural networks as unbiased function approximants make them particularly suited for this task. An algorithm that implements this idea is constructed, as a straightforward application of the likelihood-ratio hypothesis test. The algorithm compares observations with an auxiliary set of reference-distributed events, possibly obtained with a Monte Carlo event generator. It returns a p-value, which measures the compatibility of the reference model with the data. It also identifies the most discrepant phase-space region of the data set, to be selected for further investigation. The most interesting potential applications are model-independent new physics searches, although our approach could also be used to compare the theoretical predictions of different Monte Carlo event generators, or for data validation algorithms. In this work we study the performance of our algorithm on a few simple examples. The results confirm the model-independence of the approach, namely that it displays good sensitivity to a variety of putative signals. Furthermore, we show that the reach does not depend much on whether a favorable signal region is selected based on prior expectations. We identify directions for improvement towards applications to real experimental data sets.",
"Reliable data quality monitoring is a key asset in delivering collision data suitable for physics analysis in any modern large-scale high energy physics experiment. This paper focuses on the use of artificial neural networks for supervised and semi-supervised problems related to the identification of anomalies in the data collected by the CMS muon detectors. We use deep neural networks to analyze LHC collision data, represented as images organized geographically. We train a classifier capable of detecting the known anomalous behaviors with unprecedented efficiency and explore the usage of convolutional autoencoders to extend anomaly detection capabilities to unforeseen failure modes. A generalization of this strategy could pave the way to the automation of the data quality assessment process for present and future high energy physics experiments.",
"We introduce a potentially powerful new method of searching for new physics at the LHC, using autoencoders and unsupervised deep learning. The key idea of the autoencoder is that it learns to map \"normal\" events back to themselves, but fails to reconstruct \"anomalous\" events that it has never encountered before. The reconstruction error can then be used as an anomaly threshold. We demonstrate the effectiveness of this idea using QCD jets as background and boosted top jets and RPV gluino jets as signal. We show that a deep autoencoder can significantly improve signal over background when trained on backgrounds only, or even directly on data which contains a small admixture of signal. Finally we examine the correlation of the autoencoders with jet mass and show how the jet mass distribution can be stable against cuts in reconstruction loss. This may be important for estimating QCD backgrounds from data. As a test case we show how one could plausibly discover 400 GeV RPV gluinos using an autoencoder combined with a bump hunt in jet mass. This opens up the exciting possibility of training directly on actual data to discover new physics with no prior expectations or theory prejudice."
]
}
|
1811.10111
|
2901710322
|
We present the first real-time sleep staging system that uses deep learning without the need for servers in a smartphone application for a wearable EEG. We employ real-time adaptation of a single channel Electroencephalography (EEG) to infer from a Time-Distributed 1-D Deep Convolutional Neural Network. Polysomnography (PSG)-the gold standard for sleep staging, requires a human scorer and is both complex and resource-intensive. Our work demonstrates an end-to-end on-smartphone pipeline that can infer sleep stages in just single 30-second epochs, with an overall accuracy of 83.5 on 20-fold cross validation for five-class classification of sleep stages using the open Sleep-EDF dataset.
|
Automatic analysis and sleep scoring using multi-layer Neural Networks @cite_12 was done as early as 1996 using 3 channels of physiological data, namely EEG, EOG and EMG. This involved power spectral density calculations for feature extraction from raw EEG which required a tedious laboratory setting to collect reliable data through these channels. More recent work has looked into creating portable sleep scoring systems, such as the work by @cite_4 , that uses pulse, blood oxygen and motion sensors to predict sleep stages. In their paper, they do not detect sleep stages N1 and N2 separately, and N1 is usually the hardest one to predict. The authors already mention that these results cannot provide equally high accuracy as compared to the EEG and EOG signals of PSG. The same limitations apply to the work by @cite_24 . Our work achieves reliable accuracy by using only one channel from a wearable EEG, and overcomes the complexity of recording multiple signals.
|
{
"cite_N": [
"@cite_24",
"@cite_4",
"@cite_12"
],
"mid": [
"2740222873",
"2144557486",
"2412974818"
],
"abstract": [
"We focus on predicting sleep stages from radio measurements without any attached sensors on subjects. We introduce a new predictive model that combines convolutional and recurrent neural networks to extract sleep-specific subject-invariant features from RF signals and capture the temporal progression of sleep. A key innovation underlying our approach is a modified adversarial training regime that discards extraneous information specific to individuals or measurement conditions, while retaining all information relevant to the predictive task. We analyze our game theoretic setup and empirically demonstrate that our model achieves significant improvements over state-of-the-art solutions.",
"It is a well known fact that the quality of sleep is an important factor in health-related quality of life (HRQoL), and people could prevent potential problems by tracking the quality of their sleep. Unfortunately sleep scoring, which is a systematic way to address the sleep staging as well as the scoring of arousals, respiratory, cardiac, and movement events, is usually conducted with specialized equipment which is expensive and operated by specialists in dedicated sleep centers. Related research studies and products (e.g. ZEO) tried to solve this problem, but they either used multiple probes that cause discomfort to the patient, or could not score in real time. In this paper, we design and implement RASS, a portable Real-time Automatic Sleep Scoring system. RASS only requires one probe, which is inexpensive and, as a result, may be used at home or during travel. RASS accurately scores the sleeping state and detects sleep apnea in real-time based on the sensing results of pulse, blood oxygen, activity, sound and light signals. An alarm will be generated when a severely abnormal sleep state is detected. RASS has been tested with 48 patients, and the test results show that RASS could achieve higher than 84 accuracy.",
"In this paper, we compare and analyze the results from automatic analysis and visual scoring of nocturnal sleep recordings. The validation is based on a sleep recording set of 60 subjects (33 males and 27 females), consisting of three groups : 20 normal control subjects, 20 depressed patients and 20 insomniac patients treated with a benzodiazepine. The inter-expert variability estimated from these 60 recordings (61,949 epochs) indicated an average agreement rate of 87.5 between two experts on the basis of 30-second epochs. The automatic scoring system, compared in the same way with one expert, achieved an average agreement rate of 82.3 , without expert supervision. By adding expert supervision for ambiguous and unknown epochs, detected by computation of an uncertainty index and unknown rejection, the automatic expert agreement grew from 82.3 to 90 , with supervision over only 20 of the night. Bearing in mind the composition and the size of the test sample, the automated sleep staging system achieved a satisfactory performance level and may be considered a useful alternative to visual sleep stage scoring for large-scale investigations of human sleep."
]
}
|
1811.10111
|
2901710322
|
We present the first real-time sleep staging system that uses deep learning without the need for servers in a smartphone application for a wearable EEG. We employ real-time adaptation of a single channel Electroencephalography (EEG) to infer from a Time-Distributed 1-D Deep Convolutional Neural Network. Polysomnography (PSG)-the gold standard for sleep staging, requires a human scorer and is both complex and resource-intensive. Our work demonstrates an end-to-end on-smartphone pipeline that can infer sleep stages in just single 30-second epochs, with an overall accuracy of 83.5 on 20-fold cross validation for five-class classification of sleep stages using the open Sleep-EDF dataset.
|
State-of-the art network model SeqSleepNet @cite_17 processes multiple epochs and outputs the sleep labels all at once using end-to-end Hierarchical Recurrent Neural Networks. This uses all 3 channels namely, EEG, EMG and EOG in order to give the best overall accuracy of 87.1 Flexibility of this wearable also makes it more preferable than the bulky system used by @cite_18 . The smart-phone based nature of our sleep-staging application overcomes the need for a client-server architecture as used in Dreem headband @cite_0 . Our TensorFlow-Lite mobile application can also be adapted to other types of EEG devices for real-time settings.
|
{
"cite_N": [
"@cite_0",
"@cite_18",
"@cite_17"
],
"mid": [
"1983256092",
"2418216612",
"2893892260"
],
"abstract": [
"Summary Manual processing of sleep recordings is extremely time-consuming. Efforts to automate this process have shown promising results, but automatic systems are generally evaluated on private databases, not allowing accurate cross-validation with other systems. In lacking a common benchmark, the relative performances of different systems are not compared easily and advances are compromised. To address this fundamental methodological impediment to sleep study, we propose an open-access database of polysomnographic biosignals. To build this database, whole-night recordings from 200 participants [97 males (aged 42.9 ± 19.8 years) and 103 females (aged 38.3 ± 18.9 years); age range: 18–76 years] were pooled from eight different research protocols performed in three different hospital-based sleep laboratories. All recordings feature a sampling frequency of 256 Hz and an electroencephalography (EEG) montage of 4–20 channels plus standard electro-oculography (EOG), electromyography (EMG), electrocardiography (ECG) and respiratory signals. Access to the database can be obtained through the Montreal Archive of Sleep Studies (MASS) website (http: www.ceams-carsm.ca en MASS), and requires only affiliation with a research institution and prior approval by the applicant's local ethical review board. Providing the research community with access to this free and open sleep database is expected to facilitate the development and cross-validation of sleep analysis automation systems. It is also expected that such a shared resource will be a catalyst for cross-centre collaborations on difficult topics such as improving inter-rater agreement on sleep stage scoring.",
"Summary An accurate home sleep study to assess electroencephalography (EEG)-based sleep stages and EEG power would be advantageous for both clinical and research purposes, such as for longitudinal studies measuring changes in sleep stages over time. The purpose of this study was to compare sleep scoring of a single-channel EEG recorded simultaneously on the forehead against attended polysomnography. Participants were recruited from both a clinical sleep centre and a longitudinal research study investigating cognitively normal ageing and Alzheimer's disease. Analysis for overall epoch-by-epoch agreement found strong and substantial agreement between the single-channel EEG compared to polysomnography (κ = 0.67). Slow wave activity in the frontal regions was also similar when comparing the single-channel EEG device to polysomnography. As expected, Stage N1 showed poor agreement (sensitivity 0.2) due to lack of occipital electrodes. Other sleep parameters, such as sleep latency and rapid eye movement (REM) onset latency, had decreased agreement. Participants with disrupted sleep consolidation, such as from obstructive sleep apnea, also had poor agreement. We suspect that disagreement in sleep parameters between the single-channel EEG and polysomnography is due partially to altered waveform morphology and or poorer signal quality in the single-channel derivation. Our results show that single-channel EEG provides comparable results to polysomnography in assessing REM, combined Stages N2 and N3 sleep and several other parameters, including frontal slow wave activity. The data establish that single-channel EEG can be a useful research tool.",
"Automatic sleep staging has been often treated as a simple classification problem that aims at determining the label of individual target polysomnography epochs one at a time. In this paper, we tackle the task as a sequence-to-sequence classification problem that receives a sequence of multiple epochs as input and classifies all of their labels at once. For this purpose, we propose a hierarchical recurrent neural network named SeqSleepNet (source code is available at http: github.com pquochuy SeqSleepNet ). At the epoch processing level, the network consists of a filterbank layer tailored to learn frequency-domain filters for preprocessing and an attention-based recurrent layer designed for short-term sequential modeling. At the sequence processing level, a recurrent layer placed on top of the learned epoch-wise features for long-term modeling of sequential epochs. The classification is then carried out on the output vectors at every time step of the top recurrent layer to produce the sequence of output labels. Despite being hierarchical, we present a strategy to train the network in an end-to-end fashion. We show that the proposed network outperforms the state-of-the-art approaches, achieving an overall accuracy, macro F1-score, and Cohen’s kappa of 87.1 , 83.3 , and 0.815 on a publicly available dataset with 200 subjects."
]
}
|
1811.10014
|
2900474539
|
The tracking-by-detection framework requires a set of positive and negative training samples to learn robust tracking models for precise localization of target objects. However, existing tracking models mostly treat different samples independently while ignores the relationship information among them. In this paper, we propose a novel structure-aware deep neural network to overcome such limitations. In particular, we construct a graph to represent the pairwise relationships among training samples, and additionally take the natural language as the supervised information to learn both feature representations and classifiers robustly. To refine the states of the target and re-track the target when it is back to view from heavy occlusion and out of view, we elaborately design a novel subnetwork to learn the target-driven visual attentions from the guidance of both visual and natural language cues. Extensive experiments on five tracking benchmark datasets validated the effectiveness of our proposed method.
|
The idea of use multi-domain layers for the training of CNN is first proposed by Nam in @cite_6 . They pretrain a CNN using a large set of videos with tracking groundtruths to obtain a generic target representation. Their network is composed of shared layers and multiple branches of domain-specific layers, where domains correspond to individual training sequences and each branch is responsible for binary classification to identify target in each domain. Their final tracking performance is indeed great and many trackers are developed based on this idea, such as BranchOut @cite_9 , Meta-tracker @cite_3 , Real-time MDNet @cite_2 . Although these trackers are all attempt to improve MDNet from different views, however, none of them consider the structure information when pretrain their models. In addition, these trackers still adopt the local search strategy which may make them sensitive to challenging factors as mentioned above. Our tracker utilizes the GCN and natural language to take the structure information into consideration and also joint use the global and local proposals for classification which make the baseline tracker more robust to challenging factors.
|
{
"cite_N": [
"@cite_9",
"@cite_3",
"@cite_6",
"@cite_2"
],
"mid": [
"2737572441",
"2783173047",
"1857884451",
""
],
"abstract": [
"We propose an extremely simple but effective regularization technique of convolutional neural networks (CNNs), referred to as BranchOut, for online ensemble tracking. Our algorithm employs a CNN for target representation, which has a common convolutional layers but has multiple branches of fully connected layers. For better regularization, a subset of branches in the CNN are selected randomly for online learning whenever target appearance models need to be updated. Each branch may have a different number of layers to maintain variable abstraction levels of target appearances. BranchOut with multi-level target representation allows us to learn robust target appearance models with diversity and handle various challenges in visual tracking problem effectively. The proposed algorithm is evaluated in standard tracking benchmarks and shows the state-of-the-art performance even without additional pretraining on external tracking sequences.",
"This paper improves state-of-the-art on-line trackers that use deep learning. Such trackers train a deep network to pick a specified object out from the background in an initial frame (initialization) and then keep training the model as tracking proceeds (updates). Our core contribution is a meta-learning-based method to adjust deep networks for tracking using off-line training. First, we learn initial parameters and per-parameter coefficients for fast online adaptation. Second, we use training signal from future frames for robustness to target appearance variations and environment changes. The resulting networks train significantly faster during the initialization, while improving robustness and accuracy. We demonstrate this approach on top of the current highest accuracy tracking approach, tracking-by-detection based MDNet and close competitor, the correlation-based CREST. Experimental results on both standard benchmarks, OTB and VOT2016, show improvements in speed, accuracy, and robustness on both trackers.",
"We propose a novel visual tracking algorithm based on the representations from a discriminatively trained Convolutional Neural Network (CNN). Our algorithm pretrains a CNN using a large set of videos with tracking groundtruths to obtain a generic target representation. Our network is composed of shared layers and multiple branches of domain-specific layers, where domains correspond to individual training sequences and each branch is responsible for binary classification to identify target in each domain. We train each domain in the network iteratively to obtain generic target representations in the shared layers. When tracking a target in a new sequence, we construct a new network by combining the shared layers in the pretrained CNN with a new binary classification layer, which is updated online. Online tracking is performed by evaluating the candidate windows randomly sampled around the previous target state. The proposed algorithm illustrates outstanding performance in existing tracking benchmarks.",
""
]
}
|
1811.10119
|
2901976130
|
Deep learning has revolutionized the ability to learn "end-to-end" autonomous vehicle control directly from raw sensory data. While there have been recent extensions to handle forms of navigation instruction, these works are unable to capture the full distribution of possible actions that could be taken and to reason about localization of the robot within the environment. In this paper, we extend end-to-end driving networks with the ability to perform point-to-point navigation as well as probabilistic localization using only noisy GPS data. We define a novel variational network capable of learning from raw camera data of the environment as well as higher level roadmaps to predict (1) a full probability distribution over the possible control commands; and (2) a deterministic control command capable of navigating on the route specified within the map. Additionally, we formulate how our model can be used to localize the robot according to correspondences between the map and the observed visual road topology, inspired by the rough localization that human drivers can perform. We test our algorithms on real-world driving data that the vehicle has never driven through before, and integrate our point-to-point navigation algorithms onboard a full-scale autonomous vehicle for real-time performance. Our localization algorithm is also evaluated over a new set of roads and intersections to demonstrates rough pose localization even in situations without any GPS prior.
|
Our work ties in to several related efforts in both control and localization. As opposed to traditional methods for autonomous driving which typically rely on distinct algorithms for localization and mapping @cite_10 @cite_16 @cite_3 , planning @cite_22 @cite_11 @cite_31 , and control @cite_30 @cite_28 , end-to-end algorithms attempt to collapse the problem (directly from raw sensory data to output control commands) into a single learned model. The ALVINN system @cite_12 originally proposed the use of multilayer perceptron to learn the direction a vehicle should steer in 1989. Recent advancements in convolutional neural networks (CNNs) have revolutionized the ability to learn driving control (i.e. steering wheel angle or road curvature) from raw imagery @cite_21 . Followup works have incorporated conditioning on additional cues @cite_20 , including mapped information @cite_32 @cite_8 . However, these works are still quite limited in that they do not capture the uncertainty of multiple outcomes, nor present the ability to reason about discrepancy between their input modalities.
|
{
"cite_N": [
"@cite_30",
"@cite_22",
"@cite_8",
"@cite_28",
"@cite_21",
"@cite_20",
"@cite_32",
"@cite_3",
"@cite_31",
"@cite_16",
"@cite_10",
"@cite_12",
"@cite_11"
],
"mid": [
"2773226774",
"2128990851",
"2887286974",
"2112930335",
"2342840547",
"2760878839",
"2766207971",
"2152671441",
"2112271657",
"2148820580",
"2118429180",
"2167224731",
"1642831299"
],
"abstract": [
"High-end vehicles are already equipped with safety systems, such as assistive braking and automatic lane following, enhancing vehicle safety. Yet, these current solutions can only help in low-complexity driving situations. In this paper, we introduce a parallel autonomy , or shared control, framework that computes safe trajectories for an automated vehicle, based on human inputs. We minimize the deviation from the human inputs while ensuring safety via a set of collision avoidance constraints. Our method achieves safe motion even in complex driving scenarios, such as those commonly encountered in an urban setting. We introduce a receding horizon planner formulated as nonlinear model predictive control (NMPC), which includes the analytic descriptions of road boundaries and the configuration and future uncertainties of other road participants. The NMPC operates over both steering and acceleration simultaneously. We introduce a nonslip model suitable for handling complex environments with dynamic obstacles, and a nonlinear combined slip vehicle model including normal load transfer capable of handling static environments. We validate the proposed approach in two complex driving scenarios. First, in an urban environment that includes a left-turn across traffic and passing on a busy street. And second, under snow conditions on a race track with sharp turns and under complex dynamic constraints. We evaluate the performance of the method with various human driving styles. We consequently observe that the method successfully avoids collisions and generates motions with minimal intervention for parallel autonomy. We note that the method can also be applied to generate safe motion for fully autonomous vehicles.",
"A new motion planning method for robots in static workspaces is presented. This method proceeds in two phases: a learning phase and a query phase. In the learning phase, a probabilistic roadmap is constructed and stored as a graph whose nodes correspond to collision-free configurations and whose edges correspond to feasible paths between these configurations. These paths are computed using a simple and fast local planner. In the query phase, any given start and goal configurations of the robot are connected to two nodes of the roadmap; the roadmap is then searched for a path joining these two nodes. The method is general and easy to implement. It can be applied to virtually any type of holonomic robot. It requires selecting certain parameters (e.g., the duration of the learning phase) whose values depend on the scene, that is the robot and its workspace. But these values turn out to be relatively easy to choose, Increased efficiency can also be achieved by tailoring some components of the method (e.g., the local planner) to the considered robots. In this paper the method is applied to planar articulated robots with many degrees of freedom. Experimental results show that path planning can be done in a fraction of a second on a contemporary workstation ( spl ap 150 MIPS), after learning for relatively short periods of time (a few dozen seconds).",
"For human drivers, having rear and side-view mirrors is vital for safe driving. They deliver a more complete view of what is happening around the car. Human drivers also heavily exploit their mental map for navigation. Nonetheless, several methods have been published that learn driving models with only a front-facing camera and without a route planner. This lack of information renders the self-driving task quite intractable. We investigate the problem in a more realistic setting, which consists of a surround-view camera system with eight cameras, a route planner, and a CAN bus reader. In particular, we develop a sensor setup that provides data for a 360-degree view of the area surrounding the vehicle, the driving route to the destination, and low-level driving maneuvers (e.g. steering angle and speed) by human drivers. With such a sensor setup we collect a new driving dataset, covering diverse driving scenarios and varying weather illumination conditions. Finally, we learn a novel driving model by integrating information from the surround-view cameras and the route planner. Two route planners are exploited: (1) by representing the planned routes on OpenStreetMap as a stack of GPS coordinates, and (2) by rendering the planned routes on TomTom Go Mobile and recording the progression into a video. Our experiments show that: (1) 360-degree surround-view cameras help avoid failures made with a single front-view camera, in particular for city driving and intersection scenarios; and (2) route planners help the driving task significantly, especially for steering angle prediction. Code, data and more visual results will be made available at http: www.vision.ee.ethz.ch heckers Drive360.",
"In this paper, a model predictive control (MPC) approach for controlling an active front steering system in an autonomous vehicle is presented. At each time step, a trajectory is assumed to be known over a finite horizon, and an MPC controller computes the front steering angle in order to follow the trajectory on slippery roads at the highest possible entry speed. We present two approaches with different computational complexities. In the first approach, we formulate the MPC problem by using a nonlinear vehicle model. The second approach is based on successive online linearization of the vehicle model. Discussions on computational complexity and performance of the two schemes are presented. The effectiveness of the proposed MPC formulation is demonstrated by simulation and experimental tests up to 21 m s on icy roads",
"We trained a convolutional neural network (CNN) to map raw pixels from a single front-facing camera directly to steering commands. This end-to-end approach proved surprisingly powerful. With minimum training data from humans the system learns to drive in traffic on local roads with or without lane markings and on highways. It also operates in areas with unclear visual guidance such as in parking lots and on unpaved roads. The system automatically learns internal representations of the necessary processing steps such as detecting useful road features with only the human steering angle as the training signal. We never explicitly trained it to detect, for example, the outline of roads. Compared to explicit decomposition of the problem, such as lane marking detection, path planning, and control, our end-to-end system optimizes all processing steps simultaneously. We argue that this will eventually lead to better performance and smaller systems. Better performance will result because the internal components self-optimize to maximize overall system performance, instead of optimizing human-selected intermediate criteria, e.g., lane detection. Such criteria understandably are selected for ease of human interpretation which doesn't automatically guarantee maximum system performance. Smaller networks are possible because the system learns to solve the problem with the minimal number of processing steps. We used an NVIDIA DevBox and Torch 7 for training and an NVIDIA DRIVE(TM) PX self-driving car computer also running Torch 7 for determining where to drive. The system operates at 30 frames per second (FPS).",
"Deep networks trained on demonstrations of human driving have learned to follow roads and avoid obstacles. However, driving policies trained via imitation learning cannot be controlled at test time. A vehicle trained end-to-end to imitate an expert cannot be guided to take a specific turn at an upcoming intersection. This limits the utility of such systems. We propose to condition imitation learning on high-level command input. At test time, the learned driving policy functions as a chauffeur that handles sensorimotor coordination but continues to respond to navigational commands. We evaluate different architectures for conditional imitation learning in vision-based driving. We conduct experiments in realistic three-dimensional simulations of urban driving and on a 1 5 scale robotic truck that is trained to drive in a residential area. Both systems drive based on visual input yet remain responsive to high-level navigational commands. The supplementary video can be viewed at this https URL",
"How can a delivery robot navigate reliably to a destination in a new office building, with minimal prior information? To tackle this challenge, this paper introduces a two-level hierarchical approach, which integrates model-free deep learning and model-based path planning. At the low level, a neural-network motion controller, called the intention-net, is trained end-to-end to provide robust local navigation. The intention-net maps images from a single monocular camera and \"intentions\" directly to robot controls. At the high level, a path planner uses a crude map, e.g., a 2-D floor plan, to compute a path from the robot's current location to the goal. The planned path provides intentions to the intention-net. Preliminary experiments suggest that the learned motion controller is robust against perceptual uncertainty and by integrating with a path planner, it generalizes effectively to new environments and goals.",
"We present a real-time algorithm which can recover the 3D trajectory of a monocular camera, moving rapidly through a previously unknown scene. Our system, which we dub MonoSLAM, is the first successful application of the SLAM methodology from mobile robotics to the \"pure vision\" domain of a single uncontrolled camera, achieving real time but drift-free performance inaccessible to structure from motion approaches. The core of the approach is the online creation of a sparse but persistent map of natural landmarks within a probabilistic framework. Our key novel contributions include an active approach to mapping and measurement, the use of a general motion model for smooth camera movement, and solutions for monocular feature initialization and feature orientation estimation. Together, these add up to an extremely efficient and robust algorithm which runs at 30 Hz with standard PC and camera hardware. This work extends the range of robotic systems in which SLAM can be usefully applied, but also opens up new areas. We present applications of MonoSLAM to real-time 3D localization and mapping for a high-performance full-size humanoid robot and live augmented reality with a hand-held camera",
"The Rapidly-exploring Random Tree (RRT) algorithm, based on incremental sampling, efficiently computes motion plans. Although the RRT algorithm quickly produces candidate feasible solutions, it tends to converge to a solution that is far from optimal. Practical applications favor “anytime” algorithms that quickly identify an initial feasible plan, then, given more computation time available during plan execution, improve the plan toward an optimal solution. This paper describes an anytime algorithm based on the RRT* which (like the RRT) finds an initial feasible solution quickly, but (unlike the RRT) almost surely converges to an optimal solution. We present two key extensions to the RRT*, committed trajectories and branch-and-bound tree adaptation, that together enable the algorithm to make more efficient use of computation time online, resulting in an anytime algorithm for real-time implementation. We evaluate the method using a series of Monte Carlo runs in a high-fidelity simulation environment, and compare the operation of the RRT and RRT* methods. We also demonstrate experimental results for an outdoor wheeled",
"The ability to simultaneously localize a robot and accurately map its surroundings is considered by many to be a key prerequisite of truly autonomous robots. However, few approaches to this problem scale up to handle the very large number of landmarks present in real environments. Kalman filter-based algorithms, for example, require time quadratic in the number of landmarks to incorporate each sensor observation. This paper presents FastSLAM, an algorithm that recursively estimates the full posterior distribution over robot pose and landmark locations, yet scales logarithmically with the number of landmarks in the map. This algorithm is based on an exact factorization of the posterior into a product of conditional landmark distributions and a distribution over robot paths. The algorithm has been run successfully on as many as 50,000 landmarks, environments far beyond the reach of previous approaches. Experimental results demonstrate the advantages and limitations of the FastSLAM algorithm on both simulated and real-world data.",
"Discusses a significant open problem in mobile robotics: simultaneous map building and localization, which the authors define as long-term globally referenced position estimation without a priori information. This problem is difficult because of the following paradox: to move precisely, a mobile robot must have an accurate environment map; however, to build an accurate map, the mobile robot's sensing locations must be known precisely. In this way, simultaneous map building and localization can be seen to present a question of 'which came first, the chicken or the egg?' (The map or the motion?) When using ultrasonic sensing, to overcome this issue the authors equip the vehicle with multiple servo-mounted sonar sensors, to provide a means in which a subset of environment features can be precisely learned from the robot's initial location and subsequently tracked to provide precise positioning. >",
"ALVINN (Autonomous Land Vehicle In a Neural Network) is a 3-layer back-propagation network designed for the task of road following. Currently ALVINN takes images from a camera and a laser range finder as input and produces as output the direction the vehicle should travel in order to follow the road. Training has been conducted using simulated road images. Successful tests on the Carnegie Mellon autonomous navigation test vehicle indicate that the network can effectively follow real roads under certain field conditions. The representation developed to perform the task differs dramatically when the network is trained under various conditions, suggesting the possibility of a novel adaptive autonomous navigation system capable of tailoring its processing to the conditions at hand.",
""
]
}
|
1811.10119
|
2901976130
|
Deep learning has revolutionized the ability to learn "end-to-end" autonomous vehicle control directly from raw sensory data. While there have been recent extensions to handle forms of navigation instruction, these works are unable to capture the full distribution of possible actions that could be taken and to reason about localization of the robot within the environment. In this paper, we extend end-to-end driving networks with the ability to perform point-to-point navigation as well as probabilistic localization using only noisy GPS data. We define a novel variational network capable of learning from raw camera data of the environment as well as higher level roadmaps to predict (1) a full probability distribution over the possible control commands; and (2) a deterministic control command capable of navigating on the route specified within the map. Additionally, we formulate how our model can be used to localize the robot according to correspondences between the map and the observed visual road topology, inspired by the rough localization that human drivers can perform. We test our algorithms on real-world driving data that the vehicle has never driven through before, and integrate our point-to-point navigation algorithms onboard a full-scale autonomous vehicle for real-time performance. Our localization algorithm is also evaluated over a new set of roads and intersections to demonstrates rough pose localization even in situations without any GPS prior.
|
A recent line of work has tied end-to-end driving networks to variational inference @cite_29 , allowing us to look at cases where multiple directions are possible, as well as reason about robustness, atypical data, and dataset normalization. Our work extends this line in that it allows us to use the same outlook and reason about maps as an additional conditioning factor.
|
{
"cite_N": [
"@cite_29"
],
"mid": [
"2899160681"
],
"abstract": [
"This paper introduces a new method for end-to-end training of deep neural networks (DNNs) and evaluates it in the context of autonomous driving. DNN training has been shown to result in high accuracy for perception to action learning given sufficient training data. However, the trained models may fail without warning in situations with insufficient or biased training data. In this paper, we propose and evaluate a novel architecture for self-supervised learning of latent variables to detect the insufficiently trained situations. Our method also addresses training data imbalance, by learning a set of underlying latent variables that characterize the training data and evaluate potential biases. We show how these latent distributions can be leveraged to adapt and accelerate the training pipeline by training on only a fraction of the total dataset. We evaluate our approach on a challenging dataset for driving. The data is collected from a full-scale autonomous vehicle. Our method provides qualitative explanation for the latent variables learned in the model. Finally, we show how our model can be additionally trained as an end-to-end controller, directly outputting a steering control command for an autonomous vehicle."
]
}
|
1811.10119
|
2901976130
|
Deep learning has revolutionized the ability to learn "end-to-end" autonomous vehicle control directly from raw sensory data. While there have been recent extensions to handle forms of navigation instruction, these works are unable to capture the full distribution of possible actions that could be taken and to reason about localization of the robot within the environment. In this paper, we extend end-to-end driving networks with the ability to perform point-to-point navigation as well as probabilistic localization using only noisy GPS data. We define a novel variational network capable of learning from raw camera data of the environment as well as higher level roadmaps to predict (1) a full probability distribution over the possible control commands; and (2) a deterministic control command capable of navigating on the route specified within the map. Additionally, we formulate how our model can be used to localize the robot according to correspondences between the map and the observed visual road topology, inspired by the rough localization that human drivers can perform. We test our algorithms on real-world driving data that the vehicle has never driven through before, and integrate our point-to-point navigation algorithms onboard a full-scale autonomous vehicle for real-time performance. Our localization algorithm is also evaluated over a new set of roads and intersections to demonstrates rough pose localization even in situations without any GPS prior.
|
Our work also relates to several research efforts in reinforcement learning in subfields such as bridging different levels of planning hierarchies @cite_14 @cite_19 , and relating to maps as agents plan and act @cite_33 @cite_5 . This work relates to a vast literature in localization and mapping @cite_10 @cite_16 , such as visual SLAM @cite_3 , @cite_24 and place recognition @cite_17 @cite_23 . However, our notion of visual matching is much more high-level, more akin to semantic visual localization and SLAM @cite_2 @cite_6 , where the semantic-level features are driving affordances.
|
{
"cite_N": [
"@cite_14",
"@cite_33",
"@cite_3",
"@cite_6",
"@cite_24",
"@cite_19",
"@cite_23",
"@cite_2",
"@cite_5",
"@cite_16",
"@cite_10",
"@cite_17"
],
"mid": [
"2109910161",
"2258731934",
"2152671441",
"2739423245",
"612478963",
"2605016475",
"2129000642",
"2963324085",
"2964220198",
"2148820580",
"2118429180",
"2038694110"
],
"abstract": [
"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.",
"We introduce the value iteration network (VIN): a fully differentiable neural network with a planning module' embedded within. VINs can learn to plan, and are suitable for predicting outcomes that involve planning-based reasoning, such as policies for reinforcement learning. Key to our approach is a novel differentiable approximation of the value-iteration algorithm, which can be represented as a convolutional neural network, and trained end-to-end using standard backpropagation. We evaluate VIN based policies on discrete and continuous path-planning domains, and on a natural-language based search task. We show that by learning an explicit planning computation, VIN policies generalize better to new, unseen domains.",
"We present a real-time algorithm which can recover the 3D trajectory of a monocular camera, moving rapidly through a previously unknown scene. Our system, which we dub MonoSLAM, is the first successful application of the SLAM methodology from mobile robotics to the \"pure vision\" domain of a single uncontrolled camera, achieving real time but drift-free performance inaccessible to structure from motion approaches. The core of the approach is the online creation of a sparse but persistent map of natural landmarks within a probabilistic framework. Our key novel contributions include an active approach to mapping and measurement, the use of a general motion model for smooth camera movement, and solutions for monocular feature initialization and feature orientation estimation. Together, these add up to an extremely efficient and robust algorithm which runs at 30 Hz with standard PC and camera hardware. This work extends the range of robotic systems in which SLAM can be usefully applied, but also opens up new areas. We present applications of MonoSLAM to real-time 3D localization and mapping for a high-performance full-size humanoid robot and live augmented reality with a hand-held camera",
"Traditional approaches to simultaneous localization and mapping (SLAM) rely on low-level geometric features such as points, lines, and planes. They are unable to assign semantic labels to landmarks observed in the environment. Furthermore, loop closure recognition based on low-level features is often viewpoint-dependent and subject to failure in ambiguous or repetitive environments. On the other hand, object recognition methods can infer landmark classes and scales, resulting in a small set of easily recognizable landmarks, ideal for view-independent unambiguous loop closure. In a map with several objects of the same class, however, a crucial data association problem exists. While data association and recognition are discrete problems usually solved using discrete inference, classical SLAM is a continuous optimization over metric information. In this paper, we formulate an optimization problem over sensor states and semantic landmark positions that integrates metric information, semantic information, and data associations, and decompose it into two interconnected problems: an estimation of discrete data association and landmark class probabilities, and a continuous optimization over the metric states. The estimated landmark and robot poses affect the association and class distributions, which in turn affect the robot-landmark pose optimization. The performance of our algorithm is demonstrated on indoor and outdoor datasets.",
"We propose a direct (feature-less) monocular SLAM algorithm which, in contrast to current state-of-the-art regarding direct methods, allows to build large-scale, consistent maps of the environment. Along with highly accurate pose estimation based on direct image alignment, the 3D environment is reconstructed in real-time as pose-graph of keyframes with associated semi-dense depth maps. These are obtained by filtering over a large number of pixelwise small-baseline stereo comparisons. The explicitly scale-drift aware formulation allows the approach to operate on challenging sequences including large variations in scene scale. Major enablers are two key novelties: (1) a novel direct tracking method which operates on ( sim (3) ), thereby explicitly detecting scale-drift, and (2) an elegant probabilistic solution to include the effect of noisy depth values into tracking. The resulting direct monocular SLAM system runs in real-time on a CPU.",
"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.",
"Recently developed Structure from Motion (SfM) reconstruction approaches enable the creation of large scale 3D models of urban scenes. These compact scene representations can then be used for accurate image-based localization, creating the need for localization approaches that are able to efficiently handle such large amounts of data. An important bottleneck is the computation of 2D-to-3D correspondences required for pose estimation. Current stateof- the-art approaches use indirect matching techniques to accelerate this search. In this paper we demonstrate that direct 2D-to-3D matching methods have a considerable potential for improving registration performance. We derive a direct matching framework based on visual vocabulary quantization and a prioritized correspondence search. Through extensive experiments, we show that our framework efficiently handles large datasets and outperforms current state-of-the-art methods.",
"Robust visual localization under a wide range of viewing conditions is a fundamental problem in computer vision. Handling the difficult cases of this problem is not only very challenging but also of high practical relevance, e.g., in the context of life-long localization for augmented reality or autonomous robots. In this paper, we propose a novel approach based on a joint 3D geometric and semantic understanding of the world, enabling it to succeed under conditions where previous approaches failed. Our method leverages a novel generative model for descriptor learning, trained on semantic scene completion as an auxiliary task. The resulting 3D descriptors are robust to missing observations by encoding high-level 3D geometric and semantic information. Experiments on several challenging large-scale localization datasets demonstrate reliable localization under extreme viewpoint, illumination, and geometry changes.",
"This paper proposes a novel deep reinforcement learning (RL) architecture, called Value Prediction Network (VPN), which integrates model-free and model-based RL methods into a single neural network. In contrast to typical model-based RL methods, VPN learns a dynamics model whose abstract states are trained to make option-conditional predictions of future values (discounted sum of rewards) rather than of future observations. Our experimental results show that VPN has several advantages over both model-free and model-based baselines in a stochastic environment where careful planning is required but building an accurate observation-prediction model is difficult. Furthermore, VPN outperforms Deep Q-Network (DQN) on several Atari games even with short-lookahead planning, demonstrating its potential as a new way of learning a good state representation.",
"The ability to simultaneously localize a robot and accurately map its surroundings is considered by many to be a key prerequisite of truly autonomous robots. However, few approaches to this problem scale up to handle the very large number of landmarks present in real environments. Kalman filter-based algorithms, for example, require time quadratic in the number of landmarks to incorporate each sensor observation. This paper presents FastSLAM, an algorithm that recursively estimates the full posterior distribution over robot pose and landmark locations, yet scales logarithmically with the number of landmarks in the map. This algorithm is based on an exact factorization of the posterior into a product of conditional landmark distributions and a distribution over robot paths. The algorithm has been run successfully on as many as 50,000 landmarks, environments far beyond the reach of previous approaches. Experimental results demonstrate the advantages and limitations of the FastSLAM algorithm on both simulated and real-world data.",
"Discusses a significant open problem in mobile robotics: simultaneous map building and localization, which the authors define as long-term globally referenced position estimation without a priori information. This problem is difficult because of the following paradox: to move precisely, a mobile robot must have an accurate environment map; however, to build an accurate map, the mobile robot's sensing locations must be known precisely. In this way, simultaneous map building and localization can be seen to present a question of 'which came first, the chicken or the egg?' (The map or the motion?) When using ultrasonic sensing, to overcome this issue the authors equip the vehicle with multiple servo-mounted sonar sensors, to provide a means in which a subset of environment features can be precisely learned from the robot's initial location and subsequently tracked to provide precise positioning. >",
"This paper describes a probabilistic framework for appearance based navigation and mapping using spatial and visual appearance data. Like much recent work on appearance based navigation we adopt a bag-of-words approach in which positive or negative observations of visual words in a scene are used to discriminate between already visited and new places. In this paper we add an important extra dimension to the approach. We explicitly model the spatial distribution of visual words as a random graph in which nodes are visual words and edges are distributions over distances. Care is taken to ensure that the spatial model is able to capture the multi-modal distributions of inter-word spacing and account for sensor errors both in word detection and distances. Crucially, these inter-word distances are viewpoint invariant and collectively constitute strong place signatures and hence the impact of using both spatial and visual appearance is marked. We provide results illustrating a tremendous increase in precision-recall area compared to a state-of-the-art visual appearance only systems."
]
}
|
1811.10153
|
2939594267
|
We present a novel CNN-based image editing strategy that allows the user to change the semantic information of an image over an arbitrary region by manipulating the feature-space representation of the image in a trained GAN model. We will present two variants of our strategy: (1) spatial conditional batch normalization (sCBN), a type of conditional batch normalization with user-specifiable spatial weight maps, and (2) feature-blending, a method of directly modifying the intermediate features. Our methods can be used to edit both artificial image and real image, and they both can be used together with any GAN with conditional normalization layers. We will demonstrate the power of our method through experiments on various types of GANs trained on different datasets. Code will be available at https: github.com pfnet-research neural-collage.
|
In this section, we will briefly describe the ideas of related works on which our method is built, with a particular focus on two approaches from which we will borrow the philosophy. GAN is a deep generative framework consisting of a generator @math and a discriminator @math playing a min-max game where @math tries to transform a prior distribution , @math into the dataset distribution and @math tries to distinguish generated (fake) data from true samples @cite_13 . Thanks to the development of techniques against the training instability such as gradient penalty @cite_42 and spectral normalization @cite_31 , deep convolutional GANs @cite_29 are becoming the de facto standard for image generation tasks. GANs also excel in representation learning, and numerous studies reports the ability of GANs to capture semantic information. One can produce a sequence of semantically meaningful images by interpolating the latent variable @cite_29 @cite_44 @cite_27 @cite_14 .
|
{
"cite_N": [
"@cite_14",
"@cite_29",
"@cite_42",
"@cite_44",
"@cite_27",
"@cite_31",
"@cite_13"
],
"mid": [
"2952716587",
"2963684088",
"2962879692",
"2964144352",
"2548275288",
"2963836885",
"2963226019"
],
"abstract": [
"Despite recent progress in generative image modeling, successfully generating high-resolution, diverse samples from complex datasets such as ImageNet remains an elusive goal. To this end, we train Generative Adversarial Networks at the largest scale yet attempted, and study the instabilities specific to such scale. We find that applying orthogonal regularization to the generator renders it amenable to a simple \"truncation trick,\" allowing fine control over the trade-off between sample fidelity and variety by reducing the variance of the Generator's input. Our modifications lead to models which set the new state of the art in class-conditional image synthesis. When trained on ImageNet at 128x128 resolution, our models (BigGANs) achieve an Inception Score (IS) of 166.5 and Frechet Inception Distance (FID) of 7.4, improving over the previous best IS of 52.52 and FID of 18.6.",
"Abstract: In recent years, supervised learning with convolutional networks (CNNs) has seen huge adoption in computer vision applications. Comparatively, unsupervised learning with CNNs has received less attention. In this work we hope to help bridge the gap between the success of CNNs for supervised learning and unsupervised learning. We introduce a class of CNNs called deep convolutional generative adversarial networks (DCGANs), that have certain architectural constraints, and demonstrate that they are a strong candidate for unsupervised learning. Training on various image datasets, we show convincing evidence that our deep convolutional adversarial pair learns a hierarchy of representations from object parts to scenes in both the generator and discriminator. Additionally, we use the learned features for novel tasks - demonstrating their applicability as general image representations.",
"Generative Adversarial Networks (GANs) are powerful generative models, but suffer from training instability. The recently proposed Wasserstein GAN (WGAN) makes progress toward stable training of GANs, but sometimes can still generate only poor samples or fail to converge. We find that these problems are often due to the use of weight clipping in WGAN to enforce a Lipschitz constraint on the critic, which can lead to undesired behavior. We propose an alternative to clipping weights: penalize the norm of gradient of the critic with respect to its input. Our proposed method performs better than standard WGAN and enables stable training of a wide variety of GAN architectures with almost no hyperparameter tuning, including 101-layer ResNets and language models with continuous generators. We also achieve high quality generations on CIFAR-10 and LSUN bedrooms.",
"The increasingly photorealistic sample quality of generative image models suggests their feasibility in applications beyond image generation. We present the Neural Photo Editor, an interface that leverages the power of generative neural networks to make large, semantically coherent changes to existing images. To tackle the challenge of achieving accurate reconstructions without loss of feature quality, we introduce the Introspective Adversarial Network, a novel hybridization of the VAE and GAN. Our model efficiently captures long-range dependencies through use of a computational block based on weight-shared dilated convolutions, and improves generalization performance with Orthogonal Regularization, a novel weight regularization method. We validate our contributions on CelebA, SVHN, and CIFAR-100, and produce samples and reconstructions with high visual fidelity.",
"Synthesizing high resolution photorealistic images has been a long-standing challenge in machine learning. In this paper we introduce new methods for the improved training of generative adversarial networks (GANs) for image synthesis. We construct a variant of GANs employing label conditioning that results in 128x128 resolution image samples exhibiting global coherence. We expand on previous work for image quality assessment to provide two new analyses for assessing the discriminability and diversity of samples from class-conditional image synthesis models. These analyses demonstrate that high resolution samples provide class information not present in low resolution samples. Across 1000 ImageNet classes, 128x128 samples are more than twice as discriminable as artificially resized 32x32 samples. In addition, 84.7 of the classes have samples exhibiting diversity comparable to real ImageNet data.",
"One of the challenges in the study of generative adversarial networks is the instability of its training. In this paper, we propose a novel weight normalization technique called spectral normalization to stabilize the training of the discriminator. Our new normalization technique is computationally light and easy to incorporate into existing implementations. We tested the efficacy of spectral normalization on CIFAR10, STL-10, and ILSVRC2012 dataset, and we experimentally confirmed that spectrally normalized GANs (SN-GANs) is capable of generating images of better or equal quality relative to the previous training stabilization techniques.",
"This paper describes InfoGAN, an information-theoretic extension to the Generative Adversarial Network that is able to learn disentangled representations in a completely unsupervised manner. InfoGAN is a generative adversarial network that also maximizes the mutual information between a small subset of the latent variables and the observation. We derive a lower bound of the mutual information objective that can be optimized efficiently. Specifically, InfoGAN successfully disentangles writing styles from digit shapes on the MNIST dataset, pose from lighting of 3D rendered images, and background digits from the central digit on the SVHN dataset. It also discovers visual concepts that include hair styles, presence absence of eyeglasses, and emotions on the CelebA face dataset. Experiments show that InfoGAN learns interpretable representations that are competitive with representations learned by existing supervised methods. For an up-to-date version of this paper, please see https: arxiv.org abs 1606.03657."
]
}
|
1811.10153
|
2939594267
|
We present a novel CNN-based image editing strategy that allows the user to change the semantic information of an image over an arbitrary region by manipulating the feature-space representation of the image in a trained GAN model. We will present two variants of our strategy: (1) spatial conditional batch normalization (sCBN), a type of conditional batch normalization with user-specifiable spatial weight maps, and (2) feature-blending, a method of directly modifying the intermediate features. Our methods can be used to edit both artificial image and real image, and they both can be used together with any GAN with conditional normalization layers. We will demonstrate the power of our method through experiments on various types of GANs trained on different datasets. Code will be available at https: github.com pfnet-research neural-collage.
|
Class-conditional GAN @cite_27 @cite_26 @cite_21 @cite_14 is a framework designed to learn an invariant latent representation among various classes, and it is capable of generating diverse images from a same latent code @math by changing class embedding (Figure ). The work of @cite_26 @cite_14 , in particular, succeeded in producing an impressive results by interpolating the parameters of conditional batch normalization layer, which was first introduced in @cite_28 @cite_22 . Conditional batch normalization (CBN) is mechanism that learns conditional information by separately learning condition-specific scaling parameter and shifting parameter for batch normalization. Our method extends the technique used in @cite_26 by restricting the region of interpolation to a region that corresponds to the region of interest in the pixel space. We will refer to our approach (sCBN) . Unlike the manipulation done in style transfer @cite_18 , we introduce the conditional information at multiple levels in the network, depending on the style preference of the user. As we will show, sCBN in the lower layers transforms global features, and sCBN at upper layers transforms local features.
|
{
"cite_N": [
"@cite_18",
"@cite_14",
"@cite_26",
"@cite_22",
"@cite_28",
"@cite_21",
"@cite_27"
],
"mid": [
"2611605760",
"2952716587",
"2962754210",
"2963245493",
"2963921132",
"2964074081",
"2548275288"
],
"abstract": [
"We propose a new technique for visual attribute transfer across images that may have very different appearance but have perceptually similar semantic structure. By visual attribute transfer, we mean transfer of visual information (such as color, tone, texture, and style) from one image to another. For example, one image could be that of a painting or a sketch while the other is a photo of a real scene, and both depict the same type of scene. Our technique finds semantically-meaningful dense correspondences between two input images. To accomplish this, it adapts the notion of \"image analogy\" [ 2001] with features extracted from a Deep Convolutional Neutral Network for matching; we call our technique deep image analogy. A coarse-to-fine strategy is used to compute the nearest-neighbor field for generating the results. We validate the effectiveness of our proposed method in a variety of cases, including style texture transfer, color style swap, sketch painting to photo, and time lapse.",
"Despite recent progress in generative image modeling, successfully generating high-resolution, diverse samples from complex datasets such as ImageNet remains an elusive goal. To this end, we train Generative Adversarial Networks at the largest scale yet attempted, and study the instabilities specific to such scale. We find that applying orthogonal regularization to the generator renders it amenable to a simple \"truncation trick,\" allowing fine control over the trade-off between sample fidelity and variety by reducing the variance of the Generator's input. Our modifications lead to models which set the new state of the art in class-conditional image synthesis. When trained on ImageNet at 128x128 resolution, our models (BigGANs) achieve an Inception Score (IS) of 166.5 and Frechet Inception Distance (FID) of 7.4, improving over the previous best IS of 52.52 and FID of 18.6.",
"We propose a novel, projection based way to incorporate the conditional information into the discriminator of GANs that respects the role of the conditional information in the underlining probabilistic model. This approach is in contrast with most frameworks of conditional GANs used in application today, which use the conditional information by concatenating the (embedded) conditional vector to the feature vectors. With this modification, we were able to significantly improve the quality of the class conditional image generation on ILSVRC2012 (ImageNet) dataset from the current state-of-the-art result, and we achieved this with a single pair of a discriminator and a generator. We were also able to extend the application to super-resolution and succeeded in producing highly discriminative super-resolution images. This new structure also enabled high quality category transformation based on parametric functional transformation of conditional batch normalization layers in the generator.",
"It is commonly assumed that language refers to high-level visual concepts while leaving low-level visual processing unaffected. This view dominates the current literature in computational models for language-vision tasks, where visual and linguistic inputs are mostly processed independently before being fused into a single representation. In this paper, we deviate from this classic pipeline and propose to modulate the by a linguistic input. Specifically, we introduce Conditional Batch Normalization (CBN) as an efficient mechanism to modulate convolutional feature maps by a linguistic embedding. We apply CBN to a pre-trained Residual Network (ResNet), leading to the MODulatEd ResNet ( ) architecture, and show that this significantly improves strong baselines on two visual question answering tasks. Our ablation study confirms that modulating from the early stages of the visual processing is beneficial.",
"We introduce a general-purpose conditioning method for neu-ral networks called FiLM: Feature-wise Linear Modulation. FiLM layers influence neural network computation via a simple , feature-wise affine transformation based on conditioning information. We show that FiLM layers are highly effective for visual reasoning — answering image-related questions which require a multi-step, high-level process — a task which has proven difficult for standard deep learning methods that do not explicitly model reasoning. Specifically, we show on visual reasoning tasks that FiLM layers 1) halve state-of-the-art error for the CLEVR benchmark, 2) modulate features in a coherent manner, 3) are robust to ablations and architectural modifications, and 4) generalize well to challenging, new data from few examples or even zero-shot.",
"",
"Synthesizing high resolution photorealistic images has been a long-standing challenge in machine learning. In this paper we introduce new methods for the improved training of generative adversarial networks (GANs) for image synthesis. We construct a variant of GANs employing label conditioning that results in 128x128 resolution image samples exhibiting global coherence. We expand on previous work for image quality assessment to provide two new analyses for assessing the discriminability and diversity of samples from class-conditional image synthesis models. These analyses demonstrate that high resolution samples provide class information not present in low resolution samples. Across 1000 ImageNet classes, 128x128 samples are more than twice as discriminable as artificially resized 32x32 samples. In addition, 84.7 of the classes have samples exhibiting diversity comparable to real ImageNet data."
]
}
|
1811.09780
|
2901958833
|
Existing methods for single images raindrop removal either have poor robustness or suffer from parameter burdens. In this paper, we propose a new Adjacent Aggregation Network (A^2Net) with lightweight architectures to remove raindrops from single images. Instead of directly cascading convolutional layers, we design an adjacent aggregation architecture to better fuse features for rich representations generation, which can lead to high quality images reconstruction. To further simplify the learning process, we utilize a problem-specific knowledge to force the network focus on the luminance channel in the YUV color space instead of all RGB channels. By combining adjacent aggregating operation with color space transformation, the proposed A^2Net can achieve state-of-the-art performances on raindrop removal with significant parameters reduction.
|
Model-based methods are based on physical imaging process or raindrop geometric appearance to model the distribution of raindrops. In @cite_19 , the authors attempt to model the shape of adherent raindrops by a sphere section. Furthermore, in @cite_17 , Bessel curves are used to obtain higher modeling accuracy. Since raindrops have various shapes and sizes, the above models can cover only a small portion of raindrop shapes. To simplify the raindrop removal problem, several hardware constraints, e.g., multiple cameras @cite_8 , a stereo camera system @cite_28 and pan-tilt surveillance cameras @cite_27 , are exploited. However, these methods cannot work with single cameras. In @cite_6 , the authors detect raindrops by using motion and intensity temporal derivatives of input videos. Since this method requires consecutive video frames to extract raindrop features, it is not suitable for processing single images.
|
{
"cite_N": [
"@cite_8",
"@cite_28",
"@cite_6",
"@cite_19",
"@cite_27",
"@cite_17"
],
"mid": [
"2103888133",
"2158260517",
"2177195834",
"2038133417",
"2115571019",
"1585341626"
],
"abstract": [
"In this paper, we propose a new method for the restoration of deteriorated images by using multiple cameras. In outdoor environment, it is often the case that scenes taken by the cameras are hard to see because of adherent noises on the surface of the lens-protecting glass of the cameras. Our proposed method analyses multiple camera images describing the same scene, and synthesizes an image in which adherent noises are eliminated.",
"In this paper, we propose a new method that can remove view-disturbing noises from stereo images. One of the thorny problems in outdoor surveillance by a camera is that adherent noises such as waterdrops on the protecting glass surface lens disturb the view from the camera. Therefore, we propose a method for removing adherent noises from stereo images taken with a stereo camera system. Our method is based on the stereo measurement and utilizes disparities between stereo image pair. Positions of noises in images can be detected by comparing disparities measured from stereo images with the distance between the stereo camera system and the glass surface. True disparities of image regions hidden by noises can be estimated from the property that disparities are generally similar with those around noises. Finally, we can remove noises from images by replacing the above regions with textures of corresponding image regions obtained by the disparity referring. Experimental results show the effectiveness of the proposed method.",
"Raindrops adhered to a windscreen or window glass can significantly degrade the visibility of a scene. Modeling, detecting and removing raindrops will, therefore, benefit many computer vision applications, particularly outdoor surveillance systems and intelligent vehicle systems. In this paper, a method that automatically detects and removes adherent raindrops is introduced. The core idea is to exploit the local spatio-temporal derivatives of raindrops. To accomplish the idea, we first model adherent raindrops using law of physics, and detect raindrops based on these models in combination with motion and intensity temporal derivatives of the input video. Having detected the raindrops, we remove them and restore the images based on an analysis that some areas of raindrops completely occludes the scene, and some other areas occlude only partially. For partially occluding areas, we restore them by retrieving as much as possible information of the scene, namely, by solving a blending function on the detected partially occluding areas using the temporal intensity derivative. For completely occluding areas, we recover them by using a video completion technique. Experimental results using various real videos show the effectiveness of our method.",
"In this paper we present a novel approach to improved image registration in rainy weather situations. To this end, we perform monocular raindrop detection in single images based on a photometric raindrop model. Our method is capable of detecting raindrops precisely, even in front of complex backgrounds. The effectiveness is demonstrated by a significant increase in image registration accuracy which also allows for successful image restoration. Experiments on video sequences taken from within a moving vehicle prove the applicability to real-world scenarios.",
"In this paper, we propose a new method that can remove view-disturbing noises from images of dynamic scenes. One of the thorny problems in outdoor surveillance by a camera is that adherent noises such as waterdrops or mud blobs on the protecting glass surface lens disturb the view from the camera. Therefore, we propose a method for removing adherent noises from images of dynamic scenes taken by changing the direction of a pan-tilt camera, which is often used for surveillance. Our method is based on the comparison of two images, a reference image and a second image taken by a different camera angle. The latter image is transformed by a projective transformation and subtracted from the reference image to extract the regions of adherent noises and moving objects. The regions of adherent noises in the reference image are identified by examining the shapes and distances of regions existing in the subtracted image. Finally, regions of adherent noises can be eliminated by merging two images. Experimental results show the effectiveness of our proposed method.",
"In this paper, we propose a novel raindrop shape model for the detection of view-disturbing, adherent raindrops on inclined surfaces. Whereas state-of-the-art techniques do not consider inclined surfaces because they assume the droplets as sphere sections with equal contact angles, our model incorporates cubic Bezier curves that provide a low dimensional and physically interpretable representation of a raindrop surface. The parameters are empirically deduced from numerous observations of different raindrop sizes and surface inclination angles. It can be easily integrated into a probabilistic framework for raindrop recognition, using geometrical optics to simulate the visual raindrop appearance. In comparison to a sphere section model, the proposed model yields an improved droplet surface accuracy up to three orders of magnitude."
]
}
|
1811.09780
|
2901958833
|
Existing methods for single images raindrop removal either have poor robustness or suffer from parameter burdens. In this paper, we propose a new Adjacent Aggregation Network (A^2Net) with lightweight architectures to remove raindrops from single images. Instead of directly cascading convolutional layers, we design an adjacent aggregation architecture to better fuse features for rich representations generation, which can lead to high quality images reconstruction. To further simplify the learning process, we utilize a problem-specific knowledge to force the network focus on the luminance channel in the YUV color space instead of all RGB channels. By combining adjacent aggregating operation with color space transformation, the proposed A^2Net can achieve state-of-the-art performances on raindrop removal with significant parameters reduction.
|
Learning-based methods use large amounts of data to learn and explore the characteristics of raindrops. In @cite_26 , the authors detects raindrops on a windshield using raindrop features learned by PCA. When objects in the background are similar to raindrops, PCA cannot effectively extract the characteristics of raindrops and causes false detection. Recently, due to the large amount of available training data and computing resources, deep learning has become the most popular learning-based methods. In @cite_16 , the authors build a 3 layers network to extract features of static raindrops and dirt spots from synthetic images. While this method works well on small and sparse rain spots, it cannot remove large and dense raindrops. Recently, an AttentiveGAN @cite_4 is proposed to simultaneously detect and remove raindrops. This method employs a recurrent network to detect raindrops and generate corresponding attention maps. These maps are further injected into the following networks to boost the reconstruction performance.
|
{
"cite_N": [
"@cite_26",
"@cite_4",
"@cite_16"
],
"mid": [
"2113968972",
"2768189935",
"2154815154"
],
"abstract": [
"We propose a weather recognition method from in-vehicle camera images that uses a subspace method to judge rainy weather by detecting raindrops on the windshield. \"Eigendrops\" represent the principal components extracted from raindrop images in the learning stage. Then the method detects raindrops by template matching. In experiments using actual video sequences, our method showed good detection ability of raindrops and promising results for rainfall judgment from detection results.",
"Raindrops adhered to a glass window or camera lens can severely hamper the visibility of a background scene and degrade an image considerably. In this paper, we address the problem by visually removing raindrops, and thus transforming a raindrop degraded image into a clean one. The problem is intractable, since first the regions occluded by raindrops are not given. Second, the information about the background scene of the occluded regions is completely lost for most part. To resolve the problem, we apply an attentive generative network using adversarial training. Our main idea is to inject visual attention into both the generative and discriminative networks. During the training, our visual attention learns about raindrop regions and their surroundings. Hence, by injecting this information, the generative network will pay more attention to the raindrop regions and the surrounding structures, and the discriminative network will be able to assess the local consistency of the restored regions. This injection of visual attention to both generative and discriminative networks is the main contribution of this paper. Our experiments show the effectiveness of our approach, which outperforms the state of the art methods quantitatively and qualitatively.",
"Photographs taken through a window are often compromised by dirt or rain present on the window surface. Common cases of this include pictures taken from inside a vehicle, or outdoor security cameras mounted inside a protective enclosure. At capture time, defocus can be used to remove the artifacts, but this relies on achieving a shallow depth-of-field and placement of the camera close to the window. Instead, we present a post-capture image processing solution that can remove localized rain and dirt artifacts from a single image. We collect a dataset of clean corrupted image pairs which are then used to train a specialized form of convolutional neural network. This learns how to map corrupted image patches to clean ones, implicitly capturing the characteristic appearance of dirt and water droplets in natural images. Our models demonstrate effective removal of dirt and rain in outdoor test conditions."
]
}
|
1811.09780
|
2901958833
|
Existing methods for single images raindrop removal either have poor robustness or suffer from parameter burdens. In this paper, we propose a new Adjacent Aggregation Network (A^2Net) with lightweight architectures to remove raindrops from single images. Instead of directly cascading convolutional layers, we design an adjacent aggregation architecture to better fuse features for rich representations generation, which can lead to high quality images reconstruction. To further simplify the learning process, we utilize a problem-specific knowledge to force the network focus on the luminance channel in the YUV color space instead of all RGB channels. By combining adjacent aggregating operation with color space transformation, the proposed A^2Net can achieve state-of-the-art performances on raindrop removal with significant parameters reduction.
|
Since raindrops contain various shapes, sizes and appearances, most existing methods cannot simultaneously achieve performance and robustness. The recent AttentiveGAN @cite_4 can well remove raindrops from single images, as shown in Figure (c). However, this network contains a relatively large number of parameters and requires complex recurrent operations, which limits its potential value in practical applications with limited computing resources.
|
{
"cite_N": [
"@cite_4"
],
"mid": [
"2768189935"
],
"abstract": [
"Raindrops adhered to a glass window or camera lens can severely hamper the visibility of a background scene and degrade an image considerably. In this paper, we address the problem by visually removing raindrops, and thus transforming a raindrop degraded image into a clean one. The problem is intractable, since first the regions occluded by raindrops are not given. Second, the information about the background scene of the occluded regions is completely lost for most part. To resolve the problem, we apply an attentive generative network using adversarial training. Our main idea is to inject visual attention into both the generative and discriminative networks. During the training, our visual attention learns about raindrop regions and their surroundings. Hence, by injecting this information, the generative network will pay more attention to the raindrop regions and the surrounding structures, and the discriminative network will be able to assess the local consistency of the restored regions. This injection of visual attention to both generative and discriminative networks is the main contribution of this paper. Our experiments show the effectiveness of our approach, which outperforms the state of the art methods quantitatively and qualitatively."
]
}
|
1811.09889
|
2951108222
|
An end-to-end trainable ConvNet architecture, that learns to harness the power of shape representation for matching disparate image pairs, is proposed. Disparate image pairs are deemed those that exhibit strong affine variations in scale, viewpoint and projection parameters accompanied by the presence of partial or complete occlusion of objects and extreme variations in ambient illumination. Under these challenging conditions, neither local nor global feature-based image matching methods, when used in isolation, have been observed to be effective. The proposed correspondence determination scheme for matching disparate images exploits high-level shape cues that are derived from low-level local feature descriptors, thus combining the best of both worlds. A graph-based representation for the disparate image pair is generated by constructing an affinity matrix that embeds the distances between feature points in two images, thus modeling the correspondence determination problem as one of graph matching. The eigenspectrum of the affinity matrix, i.e., the learned global shape representation, is then used to further regress the transformation or homography that defines the correspondence between the source image and target image. The proposed scheme is shown to yield state-of-the-art results for both, coarse-level shape matching as well as fine point-wise correspondence determination.
|
Image matching techniques in the research literature can be broadly categorized as global shape-based techniques @cite_1 @cite_7 @cite_0 or local point-based @cite_19 @cite_17 techniques. Global shape-based matching techniques rely on extracting the overall shape of an object (or structure) within the image. High-level shape cues are extracted from the underlying images to compute a degree of similarity. Global shape-based matching techniques are further subclassified as region-based @cite_29 @cite_24 or contour-based @cite_7 @cite_0 @cite_33 . Contour-based methods exploit peripheral information to augment the underlying shape-based features. Shape skeleton-based contour matching @cite_7 and dynamic programming @cite_35 are used to compute a similarity measure such as shape context @cite_1 , chamfer distance @cite_33 or set matching-based contour similarity @cite_0 . Other contour-based matching methods include segment-based matching techniques such as hierarchical Procrustes matching @cite_23 , shape tree-based matching @cite_31 and triangle area-based matching @cite_16 . However, the aforementioned contour-based matching techniques fall short when dealing with significant articulations in the object shapes. Global region-based matching approaches characterize the underlying shapes using global descriptors such as Zernike moments @cite_29 which are invariant to affine transformations. Skeleton-based shape descriptors @cite_27 @cite_24 are better at capturing shape articulations, but their performance diminishes with increasing shape complexity.
|
{
"cite_N": [
"@cite_35",
"@cite_33",
"@cite_7",
"@cite_29",
"@cite_1",
"@cite_0",
"@cite_19",
"@cite_24",
"@cite_27",
"@cite_23",
"@cite_31",
"@cite_16",
"@cite_17"
],
"mid": [
"2097459996",
"2108927102",
"2104093257",
"2019894188",
"2057175746",
"2168709984",
"",
"2114766304",
"1875950222",
"2117564239",
"2125310690",
"1989443316",
"2166160268"
],
"abstract": [
"We present an efficient multi stage approach to detection of deformable objects in real, cluttered images given a single or few hand drawn examples as models. The method handles deformations of the object by first breaking the given model into segments at high curvature points. We allow bending at these points as it has been studied that deformation typically happens at high curvature points. The broken segments are then scaled, rotated, deformed and searched independently in the gradient image. Point maps are generated for each segment that represent the locations of the matches for that segment. We then group kpoints from the point maps of kadjacent segments using a cost function that takes into account local scale variations as well as inter-segment orientations. These matched groups yield plausible locations for the objects. In the fine matching stage, the entire object contour in the localized regions is built from the k-segment groups and given a comprehensive score in a method that uses dynamic programming. An evaluation of our algorithm on a standard dataset yielded results that are better than published work on the same dataset. At the same time, we also evaluate our algorithm on additional images with considerable object deformations to verify the robustness of our method.",
"The objective of this work is the detection of object classes, such as airplanes or horses. Instead of using a model based on salient image fragments, we show that object class detection is also possible using only the object's boundary. To this end, we develop a novel learning technique to extract class-discriminative boundary fragments. In addition to their shape, these “codebook” entries also determine the object's centroid (in the manner of [19]). Boosting is used to select discriminative combinations of boundary fragments (weak detectors) to form a strong “Boundary-Fragment-Model” (BFM) detector. The generative aspect of the model is used to determine an approximate segmentation. We demonstrate the following results: (i) the BFM detector is able to represent and detect object classes principally defined by their shape, rather than their appearance; and (ii) in comparison with other published results on several object classes (airplanes, cars-rear, cows) the BFM detector is able to exceed previous performances, and to achieve this with less supervision (such as the number of training images).",
"In this paper, we introduce a new skeleton pruning method based on contour partitioning. Any contour partition can be used, but the partitions obtained by discrete curve evolution (DCE) yield excellent results. The theoretical properties and the experiments presented demonstrate that obtained skeletons are in accord with human visual perception and stable, even in the presence of significant noise and shape variations, and have the same topology as the original skeletons. In particular, we have proven that the proposed approach never produces spurious branches, which are common when using the known skeleton pruning methods. Moreover, the proposed pruning method does not displace the skeleton points. Consequently, all skeleton points are centers of maximal disks. Again, many existing methods displace skeleton points in order to produces pruned skeletons",
"In order to retrieve an image from a large image database, the descriptor should be invariant to scale and rotation. It must also have enough discriminating power and immunity to noise for retrieval from a large image database. The Zernike moment descriptor has many desirable properties such as rotation invariance, robustness to noise, expression efficiency, fast computation and multi-level representation for describing the shapes of patterns. In this paper, we show that the Zernike moment can be used as an effective descriptor of global shape of an image in a large image database. The experimental results conducted on a database of about 6,000 images in terms of exact matching under various transformations and the similarity-based retrieval show that the proposed shape descriptor is very effective in representing shapes.",
"We present a novel approach to measuring similarity between shapes and exploit it for object recognition. In our framework, the measurement of similarity is preceded by: (1) solving for correspondences between points on the two shapes; (2) using the correspondences to estimate an aligning transform. In order to solve the correspondence problem, we attach a descriptor, the shape context, to each point. The shape context at a reference point captures the distribution of the remaining points relative to it, thus offering a globally discriminative characterization. Corresponding points on two similar shapes will have similar shape contexts, enabling us to solve for correspondences as an optimal assignment problem. Given the point correspondences, we estimate the transformation that best aligns the two shapes; regularized thin-plate splines provide a flexible class of transformation maps for this purpose. The dissimilarity between the two shapes is computed as a sum of matching errors between corresponding points, together with a term measuring the magnitude of the aligning transform. We treat recognition in a nearest-neighbor classification framework as the problem of finding the stored prototype shape that is maximally similar to that in the image. Results are presented for silhouettes, trademarks, handwritten digits, and the COIL data set.",
"We introduce a shape detection framework called Contour Context Selection for detecting objects in cluttered images using only one exemplar. Shape based detection is invariant to changes of object appearance, and can reason with geometrical abstraction of the object. Our approach uses salient contours as integral tokens for shape matching. We seek a maximal, holistic matching of shapes, which checks shape features from a large spatial extent, as well as long-range contextual relationships among object parts. This amounts to finding the correct figure ground contour labeling, and optimal correspondences between control points on around contours. This removes accidental alignments and does not hallucinate objects in background clutter, without negative training examples. We formulate this task as a set-to-set contour matching problem. Naive methods would require searching over 'exponentially' many figure ground contour labelings. We simplify this task by encoding the shape descriptor algebraically in a linear form of contour figure ground variables. This allows us to use the reliable optimization technique of Linear Programming. We demonstrate our approach on the challenging task of detecting bottles, swans and other objects in cluttered images.",
"",
"This paper presents a novel framework for the recognition of objects based on their silhouettes. The main idea is to measure the distance between two shapes as the minimum extent of deformation necessary for one shape to match the other. Since the space of deformations is very high-dimensional, three steps are taken to make the search practical: 1) define an equivalence class for shapes based on shock-graph topology, 2) define an equivalence class for deformation paths based on shock-graph transitions, and 3) avoid complexity-increasing deformation paths by moving toward shock-graph degeneracy. Despite these steps, which tremendously reduce the search requirement, there still remain numerous deformation paths to consider. To that end, we employ an edit-distance algorithm for shock graphs that finds the optimal deformation path in polynomial time. The proposed approach gives intuitive correspondences for a variety of shapes and is robust in the presence of a wide range of visual transformations. The recognition rates on two distinct databases of 99 and 216 shapes each indicate highly successful within category matches (100 percent in top three matches), which render the framework potentially usable in a range of shape-based recognition applications.",
"The type of representation used in describing shape can have a significant impact on the effectiveness of a recognition strategy. Shape has been represented by its bounding curve as well as by the medial axis representation which captures the regional interaction of the boundaries. Shape matching with the former representation is achieved by curve matching, while the latter is achieved by matching skelet al graphs. We compare the effectiveness of these two methods using approaches which we have developed recently for each. The results indicate that skelet al matching involves a higher degree of computational complexity, but is better than curve matching in the presence of articulation or rearrangement of parts. However, when these variations are not present, curve matching is a better strategy due to its lower complexity and roughly equivalent recognition rate.",
"We introduce Hierarchical Procrustes Matching (HPM), a segment-based shape matching algorithm which avoids problems associated with purely global or local methods and performs well on benchmark shape retrieval tests. The simplicity of the shape representation leads to a powerful matching algorithm which incorporates intuitive ideas about the perceptual nature of shape while being computationally efficient. This includes the ability to match similar parts even when they occur at different scales or positions. While comparison of multiscale shape representations is typically based on specific features, HPM avoids the need to extract such features. The hierarchical structure of the algorithm captures the appealing notion that matching should proceed in a global to local direction.",
"We describe a new hierarchical representation for two-dimensional objects that captures shape information at multiple levels of resolution. This representation is based on a hierarchical description of an object's boundary and can be used in an elastic matching framework, both for comparing pairs of objects and for detecting objects in cluttered images. In contrast to classical elastic models, our representation explicitly captures global shape information. This leads to richer geometric models and more accurate recognition results. Our experiments demonstrate classification results that are significantly better than the current state-of-the-art in several shape datasets. We also show initial experiments in matching shapes to cluttered images.",
"In this paper, we present a shape retrieval method using triangle-area representation for nonrigid shapes with closed contours. The representation utilizes the areas of the triangles formed by the boundary points to measure the convexity concavity of each point at different scales (or triangle side lengths). This representation is effective in capturing both local and global characteristics of a shape, invariant to translation, rotation, and scaling, and robust against noise and moderate amounts of occlusion. In the matching stage, a dynamic space warping (DSW) algorithm is employed to search efficiently for the optimal (least cost) correspondence between the points of two shapes. Then, a distance is derived based on the optimal correspondence. The performance of our method is demonstrated using four standard tests on two well-known shape databases. The results show the superiority of our method over other recent methods in the literature.",
"In this paper, we propose a novel framework for contour based object detection. Compared to previous work, our contribution is three-fold. 1) A novel shape matching scheme suitable for partial matching of edge fragments. The shape descriptor has the same geometric units as shape context but our shape representation is not histogram based. 2) Grouping of partial matching hypotheses to object detection hypotheses is expressed as maximum clique inference on a weighted graph. 3) A novel local affine-transformation to utilize the holistic shape information for scoring and ranking the shape similarity hypotheses. Consequently, each detection result not only identifies the location of the target object in the image, but also provides a precise location of its contours, since we transform a complete model contour to the image. Very competitive results on ETHZ dataset, obtained in a pure shape-based framework, demonstrate that our method achieves not only accurate object detection but also precise contour localization on cluttered background."
]
}
|
1811.09889
|
2951108222
|
An end-to-end trainable ConvNet architecture, that learns to harness the power of shape representation for matching disparate image pairs, is proposed. Disparate image pairs are deemed those that exhibit strong affine variations in scale, viewpoint and projection parameters accompanied by the presence of partial or complete occlusion of objects and extreme variations in ambient illumination. Under these challenging conditions, neither local nor global feature-based image matching methods, when used in isolation, have been observed to be effective. The proposed correspondence determination scheme for matching disparate images exploits high-level shape cues that are derived from low-level local feature descriptors, thus combining the best of both worlds. A graph-based representation for the disparate image pair is generated by constructing an affinity matrix that embeds the distances between feature points in two images, thus modeling the correspondence determination problem as one of graph matching. The eigenspectrum of the affinity matrix, i.e., the learned global shape representation, is then used to further regress the transformation or homography that defines the correspondence between the source image and target image. The proposed scheme is shown to yield state-of-the-art results for both, coarse-level shape matching as well as fine point-wise correspondence determination.
|
Both, local point-based and global shape-based image matching approaches have their advantages and shortcomings. While global shape-based descriptors are well behaved and less sensitive to outliers, they, in isolation, are insufficient to compute point-wise correspondences since they do not explicitly encode keypoint information as appearance-based descriptors do @cite_8 @cite_26 . Although global shape descriptors are shown to perform well when the image pairs exhibit significant shape deformations due to changes in viewpoint, their performance suffers in the presence of strong shape articulations @cite_26 . Region-based global descriptors are also vulnerable to instances of partial occlusions. While global shape representations have the advantage of being able to employ global shape cues for matching, their computational complexity and inability to compute accurate point-wise correspondences makes them unsuitable for most practical applications that demand reliable point-wise correspondences. In contrast, local point-based descriptors, in theory, are capable of yielding more reliable keypoint correspondences, but are often plagued by noisy matches and can also prove to be computationally expensive.
|
{
"cite_N": [
"@cite_26",
"@cite_8"
],
"mid": [
"2397509712",
"2115439018"
],
"abstract": [
"A novel multi-criteria optimization framework for matching of partially visible shapes in multiple images using joint geometric graph embedding is proposed. The proposed framework achieves matching of partial shapes in images that exhibit extreme variations in scale, orientation, viewpoint and illumination and also instances of occlusion; conditions which render impractical the use of global contour-based descriptors or local pixel-level features for shape matching. The proposed technique is based on optimization of the embedding distances of geometric features obtained from the eigenspectrum of the joint image graph, coupled with regularization over values of the mean pixel intensity or histogram of oriented gradients. It is shown to obtain successfully the correspondences denoting partial shape similarities as well as correspondences between feature points in the images. A new benchmark dataset is proposed which contains disparate image pairs with extremely challenging variations in viewing conditions when compared to an existing dataset [18]. The proposed technique is shown to significantly outperform several state-of-the-art partial shape matching techniques on both datasets.",
"We address the problem of matching images with disparate appearance arising from factors like dramatic illumination (day vs. night), age (historic vs. new) and rendering style differences. The lack of local intensity or gradient patterns in these images makes the application of pixel-level descriptors like SIFT infeasible. We propose a novel formulation for detecting and matching persistent features between such images by analyzing the eigen-spectrum of the joint image graph constructed from all the pixels in the two images. We show experimental results of our approach on a public dataset of challenging image pairs and demonstrate significant performance improvements over state-of-the-art."
]
}
|
1811.09889
|
2951108222
|
An end-to-end trainable ConvNet architecture, that learns to harness the power of shape representation for matching disparate image pairs, is proposed. Disparate image pairs are deemed those that exhibit strong affine variations in scale, viewpoint and projection parameters accompanied by the presence of partial or complete occlusion of objects and extreme variations in ambient illumination. Under these challenging conditions, neither local nor global feature-based image matching methods, when used in isolation, have been observed to be effective. The proposed correspondence determination scheme for matching disparate images exploits high-level shape cues that are derived from low-level local feature descriptors, thus combining the best of both worlds. A graph-based representation for the disparate image pair is generated by constructing an affinity matrix that embeds the distances between feature points in two images, thus modeling the correspondence determination problem as one of graph matching. The eigenspectrum of the affinity matrix, i.e., the learned global shape representation, is then used to further regress the transformation or homography that defines the correspondence between the source image and target image. The proposed scheme is shown to yield state-of-the-art results for both, coarse-level shape matching as well as fine point-wise correspondence determination.
|
In recent times, several attempts have been made to leverage the representational power of deep learning methods to improve image matching @cite_3 @cite_14 @cite_13 . Zagoruyko @cite_3 have explored a variety of CNN architectures to learn a similarity function for unsupervised matching of image patches. More recently, deep learning-based architectures have been proposed to predict and identify SIFT feature-like keypoints for incorporation into traditional SfM pipelines @cite_14 @cite_30 . Simo @cite_30 learn discriminant patch representations using a Siamese CNN architecture to identify and represent keypoints. DeTone @cite_14 @cite_10 have proposed a self-supervised CNN architecture that learns keypoints from single images by warping images to known transformations, thereby rendering image pairs for supervised learning. Yi @cite_20 use SfM reconstructions for supervised learning and prediction of keypoints by a Siamese CNN.
|
{
"cite_N": [
"@cite_30",
"@cite_13",
"@cite_14",
"@cite_3",
"@cite_10",
"@cite_20"
],
"mid": [
"1869500417",
"",
"2738401084",
"2949213045",
"2775929773",
"2949896259"
],
"abstract": [
"Deep learning has revolutionalized image-level tasks such as classification, but patch-level tasks, such as correspondence, still rely on hand-crafted features, e.g. SIFT. In this paper we use Convolutional Neural Networks (CNNs) to learn discriminant patch representations and in particular train a Siamese network with pairs of (non-)corresponding patches. We deal with the large number of potential pairs with the combination of a stochastic sampling of the training set and an aggressive mining strategy biased towards patches that are hard to classify. By using the L2 distance during both training and testing we develop 128-D descriptors whose euclidean distances reflect patch similarity, and which can be used as a drop-in replacement for any task involving SIFT. We demonstrate consistent performance gains over the state of the art, and generalize well against scaling and rotation, perspective transformation, non-rigid deformation, and illumination changes. Our descriptors are efficient to compute and amenable to modern GPUs, and are publicly available.",
"",
"We present a point tracking system powered by two deep convolutional neural networks. The first network, MagicPoint, operates on single images and extracts salient 2D points. The extracted points are \"SLAM-ready\" because they are by design isolated and well-distributed throughout the image. We compare this network against classical point detectors and discover a significant performance gap in the presence of image noise. As transformation estimation is more simple when the detected points are geometrically stable, we designed a second network, MagicWarp, which operates on pairs of point images (outputs of MagicPoint), and estimates the homography that relates the inputs. This transformation engine differs from traditional approaches because it does not use local point descriptors, only point locations. Both networks are trained with simple synthetic data, alleviating the requirement of expensive external camera ground truthing and advanced graphics rendering pipelines. The system is fast and lean, easily running 30+ FPS on a single CPU.",
"In this paper we show how to learn directly from image data (i.e., without resorting to manually-designed features) a general similarity function for comparing image patches, which is a task of fundamental importance for many computer vision problems. To encode such a function, we opt for a CNN-based model that is trained to account for a wide variety of changes in image appearance. To that end, we explore and study multiple neural network architectures, which are specifically adapted to this task. We show that such an approach can significantly outperform the state-of-the-art on several problems and benchmark datasets.",
"This paper presents a self-supervised framework for training interest point detectors and descriptors suitable for a large number of multiple-view geometry problems in computer vision. As opposed to patch-based neural networks, our fully-convolutional model operates on full-sized images and jointly computes pixel-level interest point locations and associated descriptors in one forward pass. We introduce Homographic Adaptation, a multi-scale, multi-homography approach for boosting interest point detection repeatability and performing cross-domain adaptation (e.g., synthetic-to-real). Our model, when trained on the MS-COCO generic image dataset using Homographic Adaptation, is able to repeatedly detect a much richer set of interest points than the initial pre-adapted deep model and any other traditional corner detector. The final system gives rise to state-of-the-art homography estimation results on HPatches when compared to LIFT, SIFT and ORB.",
"We introduce a novel Deep Network architecture that implements the full feature point handling pipeline, that is, detection, orientation estimation, and feature description. While previous works have successfully tackled each one of these problems individually, we show how to learn to do all three in a unified manner while preserving end-to-end differentiability. We then demonstrate that our Deep pipeline outperforms state-of-the-art methods on a number of benchmark datasets, without the need of retraining."
]
}
|
1811.09889
|
2951108222
|
An end-to-end trainable ConvNet architecture, that learns to harness the power of shape representation for matching disparate image pairs, is proposed. Disparate image pairs are deemed those that exhibit strong affine variations in scale, viewpoint and projection parameters accompanied by the presence of partial or complete occlusion of objects and extreme variations in ambient illumination. Under these challenging conditions, neither local nor global feature-based image matching methods, when used in isolation, have been observed to be effective. The proposed correspondence determination scheme for matching disparate images exploits high-level shape cues that are derived from low-level local feature descriptors, thus combining the best of both worlds. A graph-based representation for the disparate image pair is generated by constructing an affinity matrix that embeds the distances between feature points in two images, thus modeling the correspondence determination problem as one of graph matching. The eigenspectrum of the affinity matrix, i.e., the learned global shape representation, is then used to further regress the transformation or homography that defines the correspondence between the source image and target image. The proposed scheme is shown to yield state-of-the-art results for both, coarse-level shape matching as well as fine point-wise correspondence determination.
|
Unlike keypoint-based approaches, optical-flow based approaches lack the ability to model shape articulations. They also require vast amounts of training data and assume temporal supervision along with prior knowledge of frame rates. In contrast, keypoints can be discovered with relatively minimal supervision @cite_3 @cite_14 @cite_20 even from single images @cite_10 . Keypoint-based approaches are capable of reliably generating fine correspondences, making them better suited for SfM pipelines. However matching of keypoints and subsequent regularization of matches are expensive and ensuring end-to-end differentiability for SfM pipelines is challenging. While both approaches learn in a data-driven manner while maintaining end-to-end differentiability, they learn to simply regress from within the image tensor space, thereby failing to exploit the underlying shape representation.
|
{
"cite_N": [
"@cite_10",
"@cite_14",
"@cite_20",
"@cite_3"
],
"mid": [
"2775929773",
"2738401084",
"2949896259",
"2949213045"
],
"abstract": [
"This paper presents a self-supervised framework for training interest point detectors and descriptors suitable for a large number of multiple-view geometry problems in computer vision. As opposed to patch-based neural networks, our fully-convolutional model operates on full-sized images and jointly computes pixel-level interest point locations and associated descriptors in one forward pass. We introduce Homographic Adaptation, a multi-scale, multi-homography approach for boosting interest point detection repeatability and performing cross-domain adaptation (e.g., synthetic-to-real). Our model, when trained on the MS-COCO generic image dataset using Homographic Adaptation, is able to repeatedly detect a much richer set of interest points than the initial pre-adapted deep model and any other traditional corner detector. The final system gives rise to state-of-the-art homography estimation results on HPatches when compared to LIFT, SIFT and ORB.",
"We present a point tracking system powered by two deep convolutional neural networks. The first network, MagicPoint, operates on single images and extracts salient 2D points. The extracted points are \"SLAM-ready\" because they are by design isolated and well-distributed throughout the image. We compare this network against classical point detectors and discover a significant performance gap in the presence of image noise. As transformation estimation is more simple when the detected points are geometrically stable, we designed a second network, MagicWarp, which operates on pairs of point images (outputs of MagicPoint), and estimates the homography that relates the inputs. This transformation engine differs from traditional approaches because it does not use local point descriptors, only point locations. Both networks are trained with simple synthetic data, alleviating the requirement of expensive external camera ground truthing and advanced graphics rendering pipelines. The system is fast and lean, easily running 30+ FPS on a single CPU.",
"We introduce a novel Deep Network architecture that implements the full feature point handling pipeline, that is, detection, orientation estimation, and feature description. While previous works have successfully tackled each one of these problems individually, we show how to learn to do all three in a unified manner while preserving end-to-end differentiability. We then demonstrate that our Deep pipeline outperforms state-of-the-art methods on a number of benchmark datasets, without the need of retraining.",
"In this paper we show how to learn directly from image data (i.e., without resorting to manually-designed features) a general similarity function for comparing image patches, which is a task of fundamental importance for many computer vision problems. To encode such a function, we opt for a CNN-based model that is trained to account for a wide variety of changes in image appearance. To that end, we explore and study multiple neural network architectures, which are specifically adapted to this task. We show that such an approach can significantly outperform the state-of-the-art on several problems and benchmark datasets."
]
}
|
1811.09889
|
2951108222
|
An end-to-end trainable ConvNet architecture, that learns to harness the power of shape representation for matching disparate image pairs, is proposed. Disparate image pairs are deemed those that exhibit strong affine variations in scale, viewpoint and projection parameters accompanied by the presence of partial or complete occlusion of objects and extreme variations in ambient illumination. Under these challenging conditions, neither local nor global feature-based image matching methods, when used in isolation, have been observed to be effective. The proposed correspondence determination scheme for matching disparate images exploits high-level shape cues that are derived from low-level local feature descriptors, thus combining the best of both worlds. A graph-based representation for the disparate image pair is generated by constructing an affinity matrix that embeds the distances between feature points in two images, thus modeling the correspondence determination problem as one of graph matching. The eigenspectrum of the affinity matrix, i.e., the learned global shape representation, is then used to further regress the transformation or homography that defines the correspondence between the source image and target image. The proposed scheme is shown to yield state-of-the-art results for both, coarse-level shape matching as well as fine point-wise correspondence determination.
|
The proposed scheme aims to overcome the above shortcomings, by integrating both global and local methods within an end-to-end trainable deep learning framework. Inspired by @cite_8 , we leverage the representational power of deep learned feature descriptors to construct a graph representation characterized by an affinity matrix. Spectral decomposition via joint diagonalization of the affinity matrix, allows a high-level shape representation based on the eigenvectors and eigenvalues of the affinity matrix. The computed shape representation is used to regress the final homography matrix via an independent CNN sub-network. The proposed scheme is shown to be very effective in matching global image descriptors as well as estimating point-wise correspondences.
|
{
"cite_N": [
"@cite_8"
],
"mid": [
"2115439018"
],
"abstract": [
"We address the problem of matching images with disparate appearance arising from factors like dramatic illumination (day vs. night), age (historic vs. new) and rendering style differences. The lack of local intensity or gradient patterns in these images makes the application of pixel-level descriptors like SIFT infeasible. We propose a novel formulation for detecting and matching persistent features between such images by analyzing the eigen-spectrum of the joint image graph constructed from all the pixels in the two images. We show experimental results of our approach on a public dataset of challenging image pairs and demonstrate significant performance improvements over state-of-the-art."
]
}
|
1811.09938
|
2901590869
|
In visual Simultaneous Localization And Mapping (SLAM), detecting loop closures has been an important but difficult task. Currently, most solutions are based on the bag-of-words approach. Yet the possibility of deep neural network application to this task has not been fully explored due to the lack of appropriate architecture design and of sufficient training data. In this paper we demonstrate the applicability of deep neural networks by addressing both issues. Specifically we show that a feature pyramid Siamese neural network can achieve state-of-the-art performance on pairwise loop closure detection. The network is trained and tested on large-scale RGB-D datasets with a novel automatic loop closure labeling algorithm. Each image pair is labelled by how much the images overlap, allowing loop closure to be computed directly rather than by labor intensive manual labeling. We present an algorithm to adopt any large-scale generic RGB-D dataset for use in training deep loop-closure networks. We show for the first time that deep neural networks are capable of detecting loop closures, and we provide a method for generating large-scale datasets for use in evaluating and training loop closure detectors.
|
The bag-of-words methodology was first proposed for text document analysis @cite_15 and was further adapted for computer vision applications @cite_29 . For image analysis, a visual analogue of a word is used in the bag-of-words model, which is based on the vector quantization process by clustering low-level visual features of local regions or points, such as color, texture, and so forth @cite_18 .
|
{
"cite_N": [
"@cite_15",
"@cite_29",
"@cite_18"
],
"mid": [
"",
"2115973703",
"2059304719"
],
"abstract": [
"",
"Thousands of images are generated every day, which implies the necessity to classify, organise and access them using an easy, faster and efficient way. Scene classification, the classification of images into semantic categories (e.g. coast, mountains and streets), is a challenging and important problem nowadays. Many different approaches concerning scene classification have been proposed in the last few years. This article presents a detailed review of some of the most commonly used scene classification approaches. Furthermore, the surveyed techniques have been tested and their accuracy evaluated. Comparative results are shown and discussed giving the advantages and disadvantages of each methodology.",
"Content-based image retrieval (CBIR) systems require users to query images by their low-level visual content; this not only makes it hard for users to formulate queries, but also can lead to unsatisfied retrieval results. To this end, image annotation was proposed. The aim of image annotation is to automatically assign keywords to images, so image retrieval users are able to query images by keywords. Image annotation can be regarded as the image classification problem: that images are represented by some low-level features and some supervised learning techniques are used to learn the mapping between low-level features and high-level concepts (i.e., class labels). One of the most widely used feature representation methods is bag-of-words (BoW). This paper reviews related works based on the issues of improving and or applying BoW for image annotation. Moreover, many recent works (from 2006 to 2012) are compared in terms of the methodology of BoW feature generation and experimental design. In addition, several different issues in using BoW are discussed, and some important issues for future research are discussed."
]
}
|
1811.09938
|
2901590869
|
In visual Simultaneous Localization And Mapping (SLAM), detecting loop closures has been an important but difficult task. Currently, most solutions are based on the bag-of-words approach. Yet the possibility of deep neural network application to this task has not been fully explored due to the lack of appropriate architecture design and of sufficient training data. In this paper we demonstrate the applicability of deep neural networks by addressing both issues. Specifically we show that a feature pyramid Siamese neural network can achieve state-of-the-art performance on pairwise loop closure detection. The network is trained and tested on large-scale RGB-D datasets with a novel automatic loop closure labeling algorithm. Each image pair is labelled by how much the images overlap, allowing loop closure to be computed directly rather than by labor intensive manual labeling. We present an algorithm to adopt any large-scale generic RGB-D dataset for use in training deep loop-closure networks. We show for the first time that deep neural networks are capable of detecting loop closures, and we provide a method for generating large-scale datasets for use in evaluating and training loop closure detectors.
|
Currently, bag-of-words approach is the state-of-the-art method for loop closure detection @cite_37 @cite_28 @cite_3 @cite_17 in which each image is represented as a histogram of word-frequency of each word present in the dictionary generated offline from a large number of images. The computation of similarity is based on comparing the histograms @cite_2 between image pairs with certain heuristics such as spatial constraint or dynamic island @cite_38 . Image pairs with high similarity are deemed as possible loop closures.
|
{
"cite_N": [
"@cite_38",
"@cite_37",
"@cite_28",
"@cite_3",
"@cite_2",
"@cite_17"
],
"mid": [
"2785901219",
"2144824356",
"1989484209",
"1612997784",
"2047463778",
"2535547924"
],
"abstract": [
"In this letter, we introduce iBoW-LCD, a novel appearance-based loop-closure detection method. The presented approach makes use of an incremental bag-of-words (BoW) scheme based on binary descriptors to retrieve previously seen similar images, avoiding any vocabulary training stage usually required by classic BoW models. In addition, to detect loop closures, iBoW-LCD builds on the concept of dynamic islands , a simple but effective mechanism to group similar images close in time, which reduces the computational times typically associated with Bayesian frameworks. Our approach is validated using several indoor and outdoor public datasets, taken under different environmental conditions, achieving a high accuracy and outperforming other state-of-the-art solutions.",
"This paper describes a probabilistic approach to the problem of recognizing places based on their appearance. The system we present is not limited to localization, but can determine that a new observation comes from a previously unseen place, and so augment its map. Effectively this is a SLAM system in the space of appearance. Our probabilistic approach allows us to explicitly account for perceptual aliasing in the environment—identical but indistinctive observations receive a low probability of having come from the same place. We achieve this by learning a generative model of place appearance. By partitioning the learning problem into two parts, new place models can be learned online from only a single observation of a place. The algorithm complexity is linear in the number of places in the map, and is particularly suitable for online loop closure detection in mobile robotics.",
"We propose a novel method for visual place recognition using bag of words obtained from accelerated segment test (FAST)+BRIEF features. For the first time, we build a vocabulary tree that discretizes a binary descriptor space and use the tree to speed up correspondences for geometrical verification. We present competitive results with no false positives in very different datasets, using exactly the same vocabulary and settings. The whole technique, including feature extraction, requires 22 ms frame in a sequence with 26 300 images that is one order of magnitude faster than previous approaches.",
"This paper presents ORB-SLAM, a feature-based monocular simultaneous localization and mapping (SLAM) system that operates in real time, in small and large indoor and outdoor environments. The system is robust to severe motion clutter, allows wide baseline loop closing and relocalization, and includes full automatic initialization. Building on excellent algorithms of recent years, we designed from scratch a novel system that uses the same features for all SLAM tasks: tracking, mapping, relocalization, and loop closing. A survival of the fittest strategy that selects the points and keyframes of the reconstruction leads to excellent robustness and generates a compact and trackable map that only grows if the scene content changes, allowing lifelong operation. We present an exhaustive evaluation in 27 sequences from the most popular datasets. ORB-SLAM achieves unprecedented performance with respect to other state-of-the-art monocular SLAM approaches. For the benefit of the community, we make the source code public.",
"This paper is concerned with the problem of keyframe detection in appearance-based visual SLAM. Appearance SLAM models a robot's environment topologically by a graph whose nodes represent strategically interesting places that have been visited by the robot and whose arcs represent spatial connectivity between these places. Specifically, we discuss and compare various methods for identifying the next location that is sufficiently different visually from the previously visited location or node in the map graph in order to decide whether a new node should be created. We survey existing techniques of keyframe detection in image retrieval and video analysis. Using experimental results obtained from visual SLAM datasets, we conclude that the feature matching method offers the best performance among five representative methods in terms of accurately measuring the amount of appearance change between robot's views and thus can serve as a simple and effective metric for detecting keyframes. This study fills an important but missing step in the current appearance SLAM research.",
"We present ORB-SLAM2, a complete simultaneous localization and mapping (SLAM) system for monocular, stereo and RGB-D cameras, including map reuse, loop closing, and relocalization capabilities. The system works in real time on standard central processing units in a wide variety of environments from small hand-held indoors sequences, to drones flying in industrial environments and cars driving around a city. Our back-end, based on bundle adjustment with monocular and stereo observations, allows for accurate trajectory estimation with metric scale. Our system includes a lightweight localization mode that leverages visual odometry tracks for unmapped regions and matches with map points that allow for zero-drift localization. The evaluation on 29 popular public sequences shows that our method achieves state-of-the-art accuracy, being in most cases the most accurate SLAM solution. We publish the source code, not only for the benefit of the SLAM community, but with the aim of being an out-of-the-box SLAM solution for researchers in other fields."
]
}
|
1811.09938
|
2901590869
|
In visual Simultaneous Localization And Mapping (SLAM), detecting loop closures has been an important but difficult task. Currently, most solutions are based on the bag-of-words approach. Yet the possibility of deep neural network application to this task has not been fully explored due to the lack of appropriate architecture design and of sufficient training data. In this paper we demonstrate the applicability of deep neural networks by addressing both issues. Specifically we show that a feature pyramid Siamese neural network can achieve state-of-the-art performance on pairwise loop closure detection. The network is trained and tested on large-scale RGB-D datasets with a novel automatic loop closure labeling algorithm. Each image pair is labelled by how much the images overlap, allowing loop closure to be computed directly rather than by labor intensive manual labeling. We present an algorithm to adopt any large-scale generic RGB-D dataset for use in training deep loop-closure networks. We show for the first time that deep neural networks are capable of detecting loop closures, and we provide a method for generating large-scale datasets for use in evaluating and training loop closure detectors.
|
Bag-of-words models, most prominently DBoW2 @cite_28 , are built on the clustering of visual features. There have been various types of feature descriptors, such as SIFT @cite_8 , SURF @cite_9 , BRIEF @cite_13 , and ORB @cite_30 . Each of these features has its own characteristics; some are invariant towards illumination or scale but complex to compute while others may be efficient but less distinctive. These hand-crafted features are manually designed, thus none of them can be robust to all application scenarios at all times. In addition, these image representations describe the local appearance of individual patches, limiting their descriptive power with respect to global descriptor methods @cite_14 @cite_34 .
|
{
"cite_N": [
"@cite_30",
"@cite_14",
"@cite_8",
"@cite_28",
"@cite_9",
"@cite_34",
"@cite_13"
],
"mid": [
"",
"2766144513",
"",
"1989484209",
"1677409904",
"2110405746",
"1491719799"
],
"abstract": [
"",
"This paper is concerned of the loop closure detection problem, which is one of the most critical parts for visual Simultaneous Localization and Mapping (SLAM) systems. Most of state-of-the-art methods use hand-crafted features and bag-of-visual-words (BoVW) to tackle this problem. Recent development in deep learning indicates that CNN features significantly outperform hand-crafted features for image representation. This advanced technology has not been fully exploited in robotics, especially in visual SLAM systems. We propose a loop closure detection method based on convolutional neural networks (CNNs). Images are fed into a pre-trained CNN model to extract features. We pre-process CNN features instead of using them directly as most of the presented approaches did before they are used to detect loops. The workflow of extracting CNN features, processing data, computing similarity score and detecting loops is presented. Finally the performance of proposed method is evaluated on several open datasets by comparing it with Fab-Map using precision-recall metric.",
"",
"We propose a novel method for visual place recognition using bag of words obtained from accelerated segment test (FAST)+BRIEF features. For the first time, we build a vocabulary tree that discretizes a binary descriptor space and use the tree to speed up correspondences for geometrical verification. We present competitive results with no false positives in very different datasets, using exactly the same vocabulary and settings. The whole technique, including feature extraction, requires 22 ms frame in a sequence with 26 300 images that is one order of magnitude faster than previous approaches.",
"In this paper, we present a novel scale- and rotation-invariant interest point detector and descriptor, coined SURF (Speeded Up Robust Features). It approximates or even outperforms previously proposed schemes with respect to repeatability, distinctiveness, and robustness, yet can be computed and compared much faster. This is achieved by relying on integral images for image convolutions; by building on the strengths of the leading existing detectors and descriptors (in casu, using a Hessian matrix-based measure for the detector, and a distribution-based descriptor); and by simplifying these methods to the essential. This leads to a combination of novel detection, description, and matching steps. The paper presents experimental results on a standard evaluation set, as well as on imagery obtained in the context of a real-life object recognition application. Both show SURF's strong performance.",
"Learning and then recognizing a route, whether travelled during the day or at night, in clear or inclement weather, and in summer or winter is a challenging task for state of the art algorithms in computer vision and robotics. In this paper, we present a new approach to visual navigation under changing conditions dubbed SeqSLAM. Instead of calculating the single location most likely given a current image, our approach calculates the best candidate matching location within every local navigation sequence. Localization is then achieved by recognizing coherent sequences of these “local best matches”. This approach removes the need for global matching performance by the vision front-end - instead it must only pick the best match within any short sequence of images. The approach is applicable over environment changes that render traditional feature-based techniques ineffective. Using two car-mounted camera datasets we demonstrate the effectiveness of the algorithm and compare it to one of the most successful feature-based SLAM algorithms, FAB-MAP. The perceptual change in the datasets is extreme; repeated traverses through environments during the day and then in the middle of the night, at times separated by months or years and in opposite seasons, and in clear weather and extremely heavy rain. While the feature-based method fails, the sequence-based algorithm is able to match trajectory segments at 100 precision with recall rates of up to 60 .",
"We propose to use binary strings as an efficient feature point descriptor, which we call BRIEF. We show that it is highly discriminative even when using relatively few bits and can be computed using simple intensity difference tests. Furthermore, the descriptor similarity can be evaluated using the Hamming distance, which is very efficient to compute, instead of the L2 norm as is usually done. As a result, BRIEF is very fast both to build and to match. We compare it against SURF and U-SURF on standard benchmarks and show that it yields a similar or better recognition performance, while running in a fraction of the time required by either."
]
}
|
1811.09938
|
2901590869
|
In visual Simultaneous Localization And Mapping (SLAM), detecting loop closures has been an important but difficult task. Currently, most solutions are based on the bag-of-words approach. Yet the possibility of deep neural network application to this task has not been fully explored due to the lack of appropriate architecture design and of sufficient training data. In this paper we demonstrate the applicability of deep neural networks by addressing both issues. Specifically we show that a feature pyramid Siamese neural network can achieve state-of-the-art performance on pairwise loop closure detection. The network is trained and tested on large-scale RGB-D datasets with a novel automatic loop closure labeling algorithm. Each image pair is labelled by how much the images overlap, allowing loop closure to be computed directly rather than by labor intensive manual labeling. We present an algorithm to adopt any large-scale generic RGB-D dataset for use in training deep loop-closure networks. We show for the first time that deep neural networks are capable of detecting loop closures, and we provide a method for generating large-scale datasets for use in evaluating and training loop closure detectors.
|
Nevertheless, SLAM systems built on them have obtained good performance both in terms of accuracy and efficiency, and the state-of-the-art performance of ORB-SLAM @cite_3 @cite_17 has made itself one of the standard algorithms.
|
{
"cite_N": [
"@cite_3",
"@cite_17"
],
"mid": [
"1612997784",
"2535547924"
],
"abstract": [
"This paper presents ORB-SLAM, a feature-based monocular simultaneous localization and mapping (SLAM) system that operates in real time, in small and large indoor and outdoor environments. The system is robust to severe motion clutter, allows wide baseline loop closing and relocalization, and includes full automatic initialization. Building on excellent algorithms of recent years, we designed from scratch a novel system that uses the same features for all SLAM tasks: tracking, mapping, relocalization, and loop closing. A survival of the fittest strategy that selects the points and keyframes of the reconstruction leads to excellent robustness and generates a compact and trackable map that only grows if the scene content changes, allowing lifelong operation. We present an exhaustive evaluation in 27 sequences from the most popular datasets. ORB-SLAM achieves unprecedented performance with respect to other state-of-the-art monocular SLAM approaches. For the benefit of the community, we make the source code public.",
"We present ORB-SLAM2, a complete simultaneous localization and mapping (SLAM) system for monocular, stereo and RGB-D cameras, including map reuse, loop closing, and relocalization capabilities. The system works in real time on standard central processing units in a wide variety of environments from small hand-held indoors sequences, to drones flying in industrial environments and cars driving around a city. Our back-end, based on bundle adjustment with monocular and stereo observations, allows for accurate trajectory estimation with metric scale. Our system includes a lightweight localization mode that leverages visual odometry tracks for unmapped regions and matches with map points that allow for zero-drift localization. The evaluation on 29 popular public sequences shows that our method achieves state-of-the-art accuracy, being in most cases the most accurate SLAM solution. We publish the source code, not only for the benefit of the SLAM community, but with the aim of being an out-of-the-box SLAM solution for researchers in other fields."
]
}
|
1811.09938
|
2901590869
|
In visual Simultaneous Localization And Mapping (SLAM), detecting loop closures has been an important but difficult task. Currently, most solutions are based on the bag-of-words approach. Yet the possibility of deep neural network application to this task has not been fully explored due to the lack of appropriate architecture design and of sufficient training data. In this paper we demonstrate the applicability of deep neural networks by addressing both issues. Specifically we show that a feature pyramid Siamese neural network can achieve state-of-the-art performance on pairwise loop closure detection. The network is trained and tested on large-scale RGB-D datasets with a novel automatic loop closure labeling algorithm. Each image pair is labelled by how much the images overlap, allowing loop closure to be computed directly rather than by labor intensive manual labeling. We present an algorithm to adopt any large-scale generic RGB-D dataset for use in training deep loop-closure networks. We show for the first time that deep neural networks are capable of detecting loop closures, and we provide a method for generating large-scale datasets for use in evaluating and training loop closure detectors.
|
Convolutional neural networks are very powerful for learning visual representation by recognizing increasingly complicated visual patterns through the stacking of convolutional layers @cite_7 . With very deep architecture design, convolutional neural networks have achieved impressive performance on classification @cite_4 @cite_35 and object detection @cite_27 @cite_5 . The ability to learn visual representations has be transferred to other tasks such as face recognition @cite_22 and fine-grained classification @cite_31 .
|
{
"cite_N": [
"@cite_35",
"@cite_4",
"@cite_22",
"@cite_7",
"@cite_27",
"@cite_5",
"@cite_31"
],
"mid": [
"2511730936",
"2949650786",
"2096733369",
"2952186574",
"",
"2570343428",
"1975517671"
],
"abstract": [
"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 .",
"Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57 error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28 relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.",
"Despite significant recent advances in the field of face recognition [10, 14, 15, 17], implementing face verification and recognition efficiently at scale presents serious challenges to current approaches. In this paper we present a system, called FaceNet, that directly learns a mapping from face images to a compact Euclidean space where distances directly correspond to a measure of face similarity. Once this space has been produced, tasks such as face recognition, verification and clustering can be easily implemented using standard techniques with FaceNet embeddings as feature vectors.",
"Large Convolutional Network models have recently demonstrated impressive classification performance on the ImageNet benchmark. However there is no clear understanding of why they perform so well, or how they might be improved. In this paper we address both issues. We introduce a novel visualization technique that gives insight into the function of intermediate feature layers and the operation of the classifier. We also perform an ablation study to discover the performance contribution from different model layers. This enables us to find model architectures that outperform Krizhevsky al on the ImageNet classification benchmark. We show our ImageNet model generalizes well to other datasets: when the softmax classifier is retrained, it convincingly beats the current state-of-the-art results on Caltech-101 and Caltech-256 datasets.",
"",
"We introduce YOLO9000, a state-of-the-art, real-time object detection system that can detect over 9000 object categories. First we propose various improvements to the YOLO detection method, both novel and drawn from prior work. The improved model, YOLOv2, is state-of-the-art on standard detection tasks like PASCAL VOC and COCO. Using a novel, multi-scale training method the same YOLOv2 model can run at varying sizes, offering an easy tradeoff between speed and accuracy. At 67 FPS, YOLOv2 gets 76.8 mAP on VOC 2007. At 40 FPS, YOLOv2 gets 78.6 mAP, outperforming state-of-the-art methods like Faster RCNN with ResNet and SSD while still running significantly faster. Finally we propose a method to jointly train on object detection and classification. Using this method we train YOLO9000 simultaneously on the COCO detection dataset and the ImageNet classification dataset. Our joint training allows YOLO9000 to predict detections for object classes that dont have labelled detection data. We validate our approach on the ImageNet detection task. YOLO9000 gets 19.7 mAP on the ImageNet detection validation set despite only having detection data for 44 of the 200 classes. On the 156 classes not in COCO, YOLO9000 gets 16.0 mAP. YOLO9000 predicts detections for more than 9000 different object categories, all in real-time.",
"Learning fine-grained image similarity is a challenging task. It needs to capture between-class and within-class image differences. This paper proposes a deep ranking model that employs deep learning techniques to learn similarity metric directly from images. It has higher learning capability than models based on hand-crafted features. A novel multiscale network structure has been developed to describe the images effectively. An efficient triplet sampling algorithm is also proposed to learn the model with distributed asynchronized stochastic gradient. Extensive experiments show that the proposed algorithm outperforms models based on hand-crafted visual features and deep classification models."
]
}
|
1811.09938
|
2901590869
|
In visual Simultaneous Localization And Mapping (SLAM), detecting loop closures has been an important but difficult task. Currently, most solutions are based on the bag-of-words approach. Yet the possibility of deep neural network application to this task has not been fully explored due to the lack of appropriate architecture design and of sufficient training data. In this paper we demonstrate the applicability of deep neural networks by addressing both issues. Specifically we show that a feature pyramid Siamese neural network can achieve state-of-the-art performance on pairwise loop closure detection. The network is trained and tested on large-scale RGB-D datasets with a novel automatic loop closure labeling algorithm. Each image pair is labelled by how much the images overlap, allowing loop closure to be computed directly rather than by labor intensive manual labeling. We present an algorithm to adopt any large-scale generic RGB-D dataset for use in training deep loop-closure networks. We show for the first time that deep neural networks are capable of detecting loop closures, and we provide a method for generating large-scale datasets for use in evaluating and training loop closure detectors.
|
The success of deep convolutional neural networks suggests its capability of learning more detailed and general representation of images. The representation can be used to accurately indicate similarity. In fact, by ranking the similarity between images in a database, deep neural networks have already been applied to image retrieval tasks @cite_23 @cite_0 .
|
{
"cite_N": [
"@cite_0",
"@cite_23"
],
"mid": [
"2963588253",
"2340690086"
],
"abstract": [
"Image descriptors based on activations of Convolutional Neural Networks (CNNs) have become dominant in image retrieval due to their discriminative power, compactness of representation, and search efficiency. Training of CNNs, either from scratch or fine-tuning, requires a large amount of annotated data, where a high quality of annotation is often crucial. In this work, we propose to fine-tune CNNs for image retrieval on a large collection of unordered images in a fully automated manner. Reconstructed 3D models obtained by the state-of-the-art retrieval and structure-from-motion methods guide the selection of the training data. We show that both hard-positive and hard-negative examples, selected by exploiting the geometry and the camera positions available from the 3D models, enhance the performance of particular-object retrieval. CNN descriptor whitening discriminatively learned from the same training data outperforms commonly used PCA whitening. We propose a novel trainable Generalized-Mean (GeM) pooling layer that generalizes max and average pooling and show that it boosts retrieval performance. Applying the proposed method to the VGG network achieves state-of-the-art performance on the standard benchmarks: Oxford Buildings, Paris, and Holidays datasets.",
"We propose a novel approach for instance-level image retrieval. It produces a global and compact fixed-length representation for each image by aggregating many region-wise descriptors. In contrast to previous works employing pre-trained deep networks as a black box to produce features, our method leverages a deep architecture trained for the specific task of image retrieval. Our contribution is twofold: (i) we leverage a ranking framework to learn convolution and projection weights that are used to build the region features; and (ii) we employ a region proposal network to learn which regions should be pooled to form the final global descriptor. We show that using clean training data is key to the success of our approach. To that aim, we use a large scale but noisy landmark dataset and develop an automatic cleaning approach. The proposed architecture produces a global image representation in a single forward pass. Our approach significantly outperforms previous approaches based on global descriptors on standard datasets. It even surpasses most prior works based on costly local descriptor indexing and spatial verification. Additional material is available at www.xrce.xerox.com Deep-Image-Retrieval."
]
}
|
1811.09938
|
2901590869
|
In visual Simultaneous Localization And Mapping (SLAM), detecting loop closures has been an important but difficult task. Currently, most solutions are based on the bag-of-words approach. Yet the possibility of deep neural network application to this task has not been fully explored due to the lack of appropriate architecture design and of sufficient training data. In this paper we demonstrate the applicability of deep neural networks by addressing both issues. Specifically we show that a feature pyramid Siamese neural network can achieve state-of-the-art performance on pairwise loop closure detection. The network is trained and tested on large-scale RGB-D datasets with a novel automatic loop closure labeling algorithm. Each image pair is labelled by how much the images overlap, allowing loop closure to be computed directly rather than by labor intensive manual labeling. We present an algorithm to adopt any large-scale generic RGB-D dataset for use in training deep loop-closure networks. We show for the first time that deep neural networks are capable of detecting loop closures, and we provide a method for generating large-scale datasets for use in evaluating and training loop closure detectors.
|
There have also been some small-scale experiments applying convolutional neural networks to loop closure detection @cite_14 @cite_33 . However, these network designs are not sufficiently utilizing the information from the environment, causing the performance to be incomparable to the state of the art from bag-of-words models. For instance, off-the-shelf usage of convolutional features did not achieve state-of-the-art performance @cite_26 @cite_23 , unless offline data whitening is applied @cite_14 which is impractical in an online procedure.
|
{
"cite_N": [
"@cite_14",
"@cite_23",
"@cite_33",
"@cite_26"
],
"mid": [
"2766144513",
"2340690086",
"2785746569",
"2953391683"
],
"abstract": [
"This paper is concerned of the loop closure detection problem, which is one of the most critical parts for visual Simultaneous Localization and Mapping (SLAM) systems. Most of state-of-the-art methods use hand-crafted features and bag-of-visual-words (BoVW) to tackle this problem. Recent development in deep learning indicates that CNN features significantly outperform hand-crafted features for image representation. This advanced technology has not been fully exploited in robotics, especially in visual SLAM systems. We propose a loop closure detection method based on convolutional neural networks (CNNs). Images are fed into a pre-trained CNN model to extract features. We pre-process CNN features instead of using them directly as most of the presented approaches did before they are used to detect loops. The workflow of extracting CNN features, processing data, computing similarity score and detecting loops is presented. Finally the performance of proposed method is evaluated on several open datasets by comparing it with Fab-Map using precision-recall metric.",
"We propose a novel approach for instance-level image retrieval. It produces a global and compact fixed-length representation for each image by aggregating many region-wise descriptors. In contrast to previous works employing pre-trained deep networks as a black box to produce features, our method leverages a deep architecture trained for the specific task of image retrieval. Our contribution is twofold: (i) we leverage a ranking framework to learn convolution and projection weights that are used to build the region features; and (ii) we employ a region proposal network to learn which regions should be pooled to form the final global descriptor. We show that using clean training data is key to the success of our approach. To that aim, we use a large scale but noisy landmark dataset and develop an automatic cleaning approach. The proposed architecture produces a global image representation in a single forward pass. Our approach significantly outperforms previous approaches based on global descriptors on standard datasets. It even surpasses most prior works based on costly local descriptor indexing and spatial verification. Additional material is available at www.xrce.xerox.com Deep-Image-Retrieval.",
"",
"Recent results indicate that the generic descriptors extracted from the convolutional neural networks are very powerful. This paper adds to the mounting evidence that this is indeed the case. We report on a series of experiments conducted for different recognition tasks using the publicly available code and model of the network which was trained to perform object classification on ILSVRC13. We use features extracted from the network as a generic image representation to tackle the diverse range of recognition tasks of object image classification, scene recognition, fine grained recognition, attribute detection and image retrieval applied to a diverse set of datasets. We selected these tasks and datasets as they gradually move further away from the original task and data the network was trained to solve. Astonishingly, we report consistent superior results compared to the highly tuned state-of-the-art systems in all the visual classification tasks on various datasets. For instance retrieval it consistently outperforms low memory footprint methods except for sculptures dataset. The results are achieved using a linear SVM classifier (or @math distance in case of retrieval) applied to a feature representation of size 4096 extracted from a layer in the net. The representations are further modified using simple augmentation techniques e.g. jittering. The results strongly suggest that features obtained from deep learning with convolutional nets should be the primary candidate in most visual recognition tasks."
]
}
|
1811.09938
|
2901590869
|
In visual Simultaneous Localization And Mapping (SLAM), detecting loop closures has been an important but difficult task. Currently, most solutions are based on the bag-of-words approach. Yet the possibility of deep neural network application to this task has not been fully explored due to the lack of appropriate architecture design and of sufficient training data. In this paper we demonstrate the applicability of deep neural networks by addressing both issues. Specifically we show that a feature pyramid Siamese neural network can achieve state-of-the-art performance on pairwise loop closure detection. The network is trained and tested on large-scale RGB-D datasets with a novel automatic loop closure labeling algorithm. Each image pair is labelled by how much the images overlap, allowing loop closure to be computed directly rather than by labor intensive manual labeling. We present an algorithm to adopt any large-scale generic RGB-D dataset for use in training deep loop-closure networks. We show for the first time that deep neural networks are capable of detecting loop closures, and we provide a method for generating large-scale datasets for use in evaluating and training loop closure detectors.
|
Furthermore, there is a serious lack of large-scale training data adequate for training deep neural networks. In order for the networks to generalize, a dataset should contain sufficiently large numbers of images from both positive cases and negative cases. Meanwhile, there should be enough difficult loop closures that do not look very similar, as well as confusing non-closure image pairs that do look similar. However, most available loop closure datasets only contain several hundreds to thousands of images and less than 10 loop closures instances @cite_37 @cite_1 , and therefore are inadequate for training.
|
{
"cite_N": [
"@cite_37",
"@cite_1"
],
"mid": [
"2144824356",
"2162536300"
],
"abstract": [
"This paper describes a probabilistic approach to the problem of recognizing places based on their appearance. The system we present is not limited to localization, but can determine that a new observation comes from a previously unseen place, and so augment its map. Effectively this is a SLAM system in the space of appearance. Our probabilistic approach allows us to explicitly account for perceptual aliasing in the environment—identical but indistinctive observations receive a low probability of having come from the same place. We achieve this by learning a generative model of place appearance. By partitioning the learning problem into two parts, new place models can be learned online from only a single observation of a place. The algorithm complexity is linear in the number of places in the map, and is particularly suitable for online loop closure detection in mobile robotics.",
"In robotic applications of visual simultaneous localization and mapping techniques, loop-closure detection and global localization are two issues that require the capacity to recognize a previously visited place from current camera measurements. We present an online method that makes it possible to detect when an image comes from an already perceived scene using local shape and color information. Our approach extends the bag-of-words method used in image classification to incremental conditions and relies on Bayesian filtering to estimate loop-closure probability. We demonstrate the efficiency of our solution by real-time loop-closure detection under strong perceptual aliasing conditions in both indoor and outdoor image sequences taken with a handheld camera."
]
}
|
1811.09938
|
2901590869
|
In visual Simultaneous Localization And Mapping (SLAM), detecting loop closures has been an important but difficult task. Currently, most solutions are based on the bag-of-words approach. Yet the possibility of deep neural network application to this task has not been fully explored due to the lack of appropriate architecture design and of sufficient training data. In this paper we demonstrate the applicability of deep neural networks by addressing both issues. Specifically we show that a feature pyramid Siamese neural network can achieve state-of-the-art performance on pairwise loop closure detection. The network is trained and tested on large-scale RGB-D datasets with a novel automatic loop closure labeling algorithm. Each image pair is labelled by how much the images overlap, allowing loop closure to be computed directly rather than by labor intensive manual labeling. We present an algorithm to adopt any large-scale generic RGB-D dataset for use in training deep loop-closure networks. We show for the first time that deep neural networks are capable of detecting loop closures, and we provide a method for generating large-scale datasets for use in evaluating and training loop closure detectors.
|
Moreover, the ground truth matrices provided in many existing datasets are usually not based on the visual similarity but on scene categories (i.e., kitchen or bedroom). Other larger image datasets do not provide the ground truth for loop closures at all @cite_10 @cite_6 . To the best of our knowledge, there is currently no proper dataset for the training of a deep neural network for loop closure application.
|
{
"cite_N": [
"@cite_10",
"@cite_6"
],
"mid": [
"2085411191",
"125693051"
],
"abstract": [
"In this paper we explore how a structured light depth sensor, in the form of the Microsoft Kinect, can assist with indoor scene segmentation. We use a CRF-based model to evaluate a range of different representations for depth information and propose a novel prior on 3D location. We introduce a new and challenging indoor scene dataset, complete with accurate depth maps and dense label coverage. Evaluating our model on this dataset reveals that the combination of depth and intensity images gives dramatic performance gains over intensity images alone. Our results clearly demonstrate the utility of structured light sensors for scene understanding.",
"We present an approach to interpret the major surfaces, objects, and support relations of an indoor scene from an RGBD image. Most existing work ignores physical interactions or is applied only to tidy rooms and hallways. Our goal is to parse typical, often messy, indoor scenes into floor, walls, supporting surfaces, and object regions, and to recover support relationships. One of our main interests is to better understand how 3D cues can best inform a structured 3D interpretation. We also contribute a novel integer programming formulation to infer physical support relations. We offer a new dataset of 1449 RGBD images, capturing 464 diverse indoor scenes, with detailed annotations. Our experiments demonstrate our ability to infer support relations in complex scenes and verify that our 3D scene cues and inferred support lead to better object segmentation."
]
}
|
1811.09938
|
2901590869
|
In visual Simultaneous Localization And Mapping (SLAM), detecting loop closures has been an important but difficult task. Currently, most solutions are based on the bag-of-words approach. Yet the possibility of deep neural network application to this task has not been fully explored due to the lack of appropriate architecture design and of sufficient training data. In this paper we demonstrate the applicability of deep neural networks by addressing both issues. Specifically we show that a feature pyramid Siamese neural network can achieve state-of-the-art performance on pairwise loop closure detection. The network is trained and tested on large-scale RGB-D datasets with a novel automatic loop closure labeling algorithm. Each image pair is labelled by how much the images overlap, allowing loop closure to be computed directly rather than by labor intensive manual labeling. We present an algorithm to adopt any large-scale generic RGB-D dataset for use in training deep loop-closure networks. We show for the first time that deep neural networks are capable of detecting loop closures, and we provide a method for generating large-scale datasets for use in evaluating and training loop closure detectors.
|
To achieve this goal and better utilize information from the environment, we add an input channel to take depth information. This provides information about the structure of the scene and is invariant to lighting conditions. The input is passed down a feature pyramid @cite_19 to capture object representations from different scales.
|
{
"cite_N": [
"@cite_19"
],
"mid": [
"2949533892"
],
"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."
]
}
|
1811.09786
|
2952164680
|
Recurrent neural networks (RNNs) such as long short-term memory and gated recurrent units are pivotal building blocks across a broad spectrum of sequence modeling problems. This paper proposes a recurrently controlled recurrent network (RCRN) for expressive and powerful sequence encoding. More concretely, the key idea behind our approach is to learn the recurrent gating functions using recurrent networks. Our architecture is split into two components - a controller cell and a listener cell whereby the recurrent controller actively influences the compositionality of the listener cell. We conduct extensive experiments on a myriad of tasks in the NLP domain such as sentiment analysis (SST, IMDb, Amazon reviews, etc.), question classification (TREC), entailment classification (SNLI, SciTail), answer selection (WikiQA, TrecQA) and reading comprehension (NarrativeQA). Across all 26 datasets, our results demonstrate that RCRN not only consistently outperforms BiLSTMs but also stacked BiLSTMs, suggesting that our controller architecture might be a suitable replacement for the widely adopted stacked architecture.
|
Another line of work is also concerned with eliminating recurrence. SRUs (Simple Recurrent Units) are recently proposed networks that remove the sequential dependencies in RNNs. SRUs can be considered a special case of Quasi-RNNs , which performs incremental pooling using pre-learned convolutional gates. A recent work, Multi-range Reasoning Units (MRU) follows the same paradigm, trading convolutional gates with features learned via expressive multi-granular reasoning. @cite_0 proposed sentence-state LSTMs (S-LSTM) that exchanges incremental reading for a single global state.
|
{
"cite_N": [
"@cite_0"
],
"mid": [
"2798915520"
],
"abstract": [
"Bi-directional LSTMs are a powerful tool for text representation. On the other hand, they have been shown to suffer various limitations due to their sequential nature. We investigate an alternative LSTM structure for encoding text, which consists of a parallel state for each word. Recurrent steps are used to perform local and global information exchange between words simultaneously, rather than incremental reading of a sequence of words. Results on various classification and sequence labelling benchmarks show that the proposed model has strong representation power, giving highly competitive performances compared to stacked BiLSTM models with similar parameter numbers."
]
}
|
1811.09852
|
2808438073
|
Program repair research has made tremendous progress over the last few years, and software development bots are now being invented to help developers gain productivity. In this paper, we investigate the concept of a "program repair bot" and present Repairnator. The Repairnator bot is an autonomous agent that constantly monitors test failures, reproduces bugs, and runs program repair tools against each reproduced bug. If a patch is found, Repairnator bot reports it to the developers. At the time of writing, Repairnator uses three different program repair systems and has been operating since February 2017. In total, it has studied 11 523 test failures over 1 609 open-source software projects hosted on GitHub, and has generated patches for 15 different bugs. Over months, we hit a number of hard technical challenges and had to make various design and engineering decisions. This gives us a unique experience in this area. In this paper, we reflect upon Repairnator in order to share this knowledge with the automatic program repair community.
|
The role of bots and how they are improving developer productivity is studied by Storey and Zagalasky @cite_25 . They provide some categories to classify the existing development bots using and criteria. They pinpoint the importance in automating tedious tasks and in keeping developers in the loop by integrating bots in developer existing environment. They also discuss the question of developers trusting bots if they generate artifacts automatically.
|
{
"cite_N": [
"@cite_25"
],
"mid": [
"2546705944"
],
"abstract": [
"Bots are used to support different software development activities, from automating repetitive tasks to bridging knowledge and communication gaps in software teams. We anticipate the use of Bots will increase and lead to improvements in software quality and developer and team productivity, but what if the disruptive effect is not what we expect? Our goal in this paper is to provoke and inspire researchers to study the impact (positive and negative) of Bots on software development. We outline the modern Bot landscape and use examples to describe the common roles Bots occupy in software teams. We propose a preliminary cognitive support framework that can be used to understand these roles and to reflect on the impact of Bots in software development on productivity. Finally, we consider challenges that Bots may bring and propose some directions for future research."
]
}
|
1811.09852
|
2808438073
|
Program repair research has made tremendous progress over the last few years, and software development bots are now being invented to help developers gain productivity. In this paper, we investigate the concept of a "program repair bot" and present Repairnator. The Repairnator bot is an autonomous agent that constantly monitors test failures, reproduces bugs, and runs program repair tools against each reproduced bug. If a patch is found, Repairnator bot reports it to the developers. At the time of writing, Repairnator uses three different program repair systems and has been operating since February 2017. In total, it has studied 11 523 test failures over 1 609 open-source software projects hosted on GitHub, and has generated patches for 15 different bugs. Over months, we hit a number of hard technical challenges and had to make various design and engineering decisions. This gives us a unique experience in this area. In this paper, we reflect upon Repairnator in order to share this knowledge with the automatic program repair community.
|
A similar problem is studied by Murgia @cite_26 . In this article, they show the compared impacts of two identical bots answering questions of developers on the Stackoverflow platform. The only difference between the two bots is their identity: the first one is presented as a human being, and the other one is clearly displayed (name, avatar) as a programmatic bot. Their results show that the developers had a really higher confidence in results provided by the human'' bot. The authors explain that developers certainly have a very low tolerance and very high expectations answers or artifacts generated by bots.
|
{
"cite_N": [
"@cite_26"
],
"mid": [
"2347054793"
],
"abstract": [
"With the rise of social media and advancements in AI technology, human-bot interaction will soon be commonplace. In this paper we explore human-bot interaction in STACK OVERFLOW, a question and answer website for developers. For this purpose, we built a bot emulating an ordinary user answering questions concerning the resolution of git error messages. In a first run this bot impersonated a human, while in a second run the same bot revealed its machine identity. Despite being functionally identical, the two bot variants elicited quite different reactions."
]
}
|
1811.09852
|
2808438073
|
Program repair research has made tremendous progress over the last few years, and software development bots are now being invented to help developers gain productivity. In this paper, we investigate the concept of a "program repair bot" and present Repairnator. The Repairnator bot is an autonomous agent that constantly monitors test failures, reproduces bugs, and runs program repair tools against each reproduced bug. If a patch is found, Repairnator bot reports it to the developers. At the time of writing, Repairnator uses three different program repair systems and has been operating since February 2017. In total, it has studied 11 523 test failures over 1 609 open-source software projects hosted on GitHub, and has generated patches for 15 different bugs. Over months, we hit a number of hard technical challenges and had to make various design and engineering decisions. This gives us a unique experience in this area. In this paper, we reflect upon Repairnator in order to share this knowledge with the automatic program repair community.
|
CCBot @cite_22 is a bot dedicated to automatically insert new contracts in C # projects. It has been created to help developers manage the results of the static analysis tools. The bot is integrated on GitHub and automatically builds projects, analyzes code contracts and proposes code changes for fixing warnings. The code changes are made as pull requests proposed to the developers. CCBot has been validated on 4 C # projects on GitHub and its authors obtained 22 merged PR.
|
{
"cite_N": [
"@cite_22"
],
"mid": [
"2548576209"
],
"abstract": [
"Existing static analysis tools require significant programmer effort. On large code bases, static analysis tools produce thousands of warnings. It is unrealistic to expect users to review such a massive list and to manually make changes for each warning. To address this issue we propose CCBot (short for C ode C ontracts Bot ), a new tool that applies the results of static analysis to existing code through automatic code transformation. Specifically, CCBot instruments the code with method preconditions, postconditions, and object invariants which detect faults at runtime or statically using a static contract checker. The only configuration the programmer needs to perform is to give CCBot the file paths to code she wants instrumented. This allows the programmer to adopt contract-based static analysis with little effort. CCBot's instrumented version of the code is guaranteed to compile if the original code did. This guarantee means the programmer can deploy or test the instrumented code immediately without additional manual effort. The inserted contracts can detect common errors such as null pointer dereferences and out-of-bounds array accesses. CCBot is a robust large-scale tool with an open-source C# implementation. We have tested it on real world projects with tens of thousands of lines of code. We discuss several projects as case studies, highlighting undiscovered bugs found by CCBot, including 22 new contracts that were accepted by the project authors."
]
}
|
1811.09852
|
2808438073
|
Program repair research has made tremendous progress over the last few years, and software development bots are now being invented to help developers gain productivity. In this paper, we investigate the concept of a "program repair bot" and present Repairnator. The Repairnator bot is an autonomous agent that constantly monitors test failures, reproduces bugs, and runs program repair tools against each reproduced bug. If a patch is found, Repairnator bot reports it to the developers. At the time of writing, Repairnator uses three different program repair systems and has been operating since February 2017. In total, it has studied 11 523 test failures over 1 609 open-source software projects hosted on GitHub, and has generated patches for 15 different bugs. Over months, we hit a number of hard technical challenges and had to make various design and engineering decisions. This gives us a unique experience in this area. In this paper, we reflect upon Repairnator in order to share this knowledge with the automatic program repair community.
|
Balachandran presents ReviewBot @cite_35 and its extension called Fix-it @cite_5 . ReviewBot'' is a standalone bot responsible for doing code review of Java programs. The bot is based on static analysis tools to detect standard code violations and common defect patterns. Then ReviewBot has been extended with Fix-it, which aims at automatically fixing some common defects identified during review. Fix-it is based on maintaining an AST of the program and performing AST transformation to fix the bad smells.
|
{
"cite_N": [
"@cite_35",
"@cite_5"
],
"mid": [
"2151979607",
"2070834360"
],
"abstract": [
"Peer code review is a cost-effective software defect detection technique. Tool assisted code review is a form of peer code review, which can improve both quality and quantity of reviews. However, there is a significant amount of human effort involved even in tool based code reviews. Using static analysis tools, it is possible to reduce the human effort by automating the checks for coding standard violations and common defect patterns. Towards this goal, we propose a tool called Review Bot for the integration of automatic static analysis with the code review process. Review Bot uses output of multiple static analysis tools to publish reviews automatically. Through a user study, we show that integrating static analysis tools with code review process can improve the quality of code review. The developer feedback for a subset of comments from automatic reviews shows that the developers agree to fix 93 of all the automatically generated comments. There is only 14.71 of all the accepted comments which need improvements in terms of priority, comment message, etc. Another problem with tool assisted code review is the assignment of appropriate reviewers. Review Bot solves this problem by generating reviewer recommendations based on change history of source code lines. Our experimental results show that the recommendation accuracy is in the range of 60 -92 , which is significantly better than a comparable method based on file change history.",
"Coding standard violations, defect patterns and non-conformance to best practices are abundant in checked-in source code. This often leads to unmaintainable code and potential bugs in later stages of software life cycle. It is important to detect and correct these issues early in the development cycle, when it is less expensive to fix. Even though static analysis techniques such as tool-assisted code review are effective in addressing this problem, there is significant amount of human effort involved in identifying the source code issues and fixing it. Review Bot is a tool designed to reduce the human effort and improve the quality in code reviews by generating automatic reviews using static analysis output. In this paper, we propose an extension to Review Bot- addition of a component called Fix-it for the auto-correction of various source code issues using Abstract Syntax Tree (AST) transformations. Fix-it uses built-in fixes to automatically fix various issues reported by the auto-reviewer component in Review Bot, thereby reducing the human effort to greater extent. Fix-it is designed to be highly extensible-users can add support for the detection of new defect patterns using XPath or XQuery and provide fixes for it based on AST transformations written in a high-level programming language. It allows the user to treat the AST as a DOM tree and run XQuery UPDATE expressions to perform AST transformations as part of a fix. Fix-it also includes a designer application which enables Review Bot administrators to design new defect patterns and fixes. The developer feedback on a stand-alone prototype indicates the possibility of significant human effort reduction in code reviews using Fix-it."
]
}
|
1811.09852
|
2808438073
|
Program repair research has made tremendous progress over the last few years, and software development bots are now being invented to help developers gain productivity. In this paper, we investigate the concept of a "program repair bot" and present Repairnator. The Repairnator bot is an autonomous agent that constantly monitors test failures, reproduces bugs, and runs program repair tools against each reproduced bug. If a patch is found, Repairnator bot reports it to the developers. At the time of writing, Repairnator uses three different program repair systems and has been operating since February 2017. In total, it has studied 11 523 test failures over 1 609 open-source software projects hosted on GitHub, and has generated patches for 15 different bugs. Over months, we hit a number of hard technical challenges and had to make various design and engineering decisions. This gives us a unique experience in this area. In this paper, we reflect upon Repairnator in order to share this knowledge with the automatic program repair community.
|
Beschastnikh @cite_13 set the concept of a common platform for software engineering research tools. They envision in their paper an ecosystem of bots dedicated to software development platform such as GitHub or Bitbucket, which would be able to submit a pull request containing a bug fix, or helping to improve the documentation.
|
{
"cite_N": [
"@cite_13"
],
"mid": [
"2732909020"
],
"abstract": [
"An important part of software engineering (SE) research is to develop new analysis techniques and to integrate these techniques into software development practice. However, since access to developers is non-trivial and research tool adoption is slow, new analyses are typically evaluated as follows: a prototype tool that embeds the analysis is implemented, a set of projects is identified, their revisions are selected, and the tool is run in a controlled environment, rarely involving the developers of the software. As a result, research artifacts are brittle and it is unclear if an analysis tool would actually be adopted. In this paper, we envision harnessing the rich interfaces provided by popular social coding platforms for automated deployment and evaluation of SE research analysis. We propose that SE analyses can be deployed as analysis bots. We focus on two specific benefits of such an approach: (1) analysis bots can help evaluate analysis techniques in a less controlled, and more realistic context, and (2) analysis bots provide an interface for developers to \"subscribe\" to new research techniques without needing to trust the implementation, the developer of the new tool, or to install the analysis tool locally. We outline basic requirements for an analysis bots platform, and present research challenges that would need to be resolved for bots to flourish."
]
}
|
1811.09675
|
2951583262
|
Dense 3D shape acquisition of swimming human or live fish is an important research topic for sports, biological science and so on. For this purpose, active stereo sensor is usually used in the air, however it cannot be applied to the underwater environment because of refraction, strong light attenuation and severe interference of bubbles. Passive stereo is a simple solution for capturing dynamic scenes at underwater environment, however the shape with textureless surfaces or irregular reflections cannot be recovered. Recently, the stereo camera pair with a pattern projector for adding artificial textures on the objects is proposed. However, to use the system for underwater environment, several problems should be compensated, i.e., disturbance by fluctuation and bubbles. Simple solution is to use convolutional neural network for stereo to cancel the effects of bubbles and or water fluctuation. Since it is not easy to train CNN with small size of database with large variation, we develop a special bubble generation device to efficiently create real bubble database of multiple size and density. In addition, we propose a transfer learning technique for multi-scale CNN to effectively remove bubbles and projected-patterns on the object. Further, we develop a real system and actually captured live swimming human, which has not been done before. Experiments are conducted to show the effectiveness of our method compared with the state of the art techniques.
|
In terms of light attenuation and disturbance problem for water medium, light transport analysis has been conducted @cite_28 @cite_13 . Narasimhan al proposed a structured-light-based 3D scanning method for strong scattering and absorption media based on light transport analysis @cite_9 . For weak scattering media, Bleier and N "uchter used cross laser projector which only achieved sparse reconstruction @cite_14 . To increase density, Campos and Codina projected parallel lines with DOE to capture underwater objects with one-shot scan @cite_2 . proposed a grid pattern to capture more dense shape with one-shot scan @cite_1 . One drawback of those one-shot scanning techniques is that reconstruction tends to be unstable even if light attenuation and disturbances are not so strong because sensitivity of pattern detection is high for subtle change of projected pattern. Some research such as @cite_34 used infrared structured light or ToF sensors, but infrared attenuates rapidly in water as shown in Fig. , and is not practical.
|
{
"cite_N": [
"@cite_14",
"@cite_28",
"@cite_9",
"@cite_1",
"@cite_2",
"@cite_34",
"@cite_13"
],
"mid": [
"2590711719",
"2327449852",
"2157932170",
"2612243049",
"2325397265",
"2743552164",
"2110637376"
],
"abstract": [
"Abstract. In-situ calibration of structured light scanners in underwater environments is time-consuming and complicated. This paper presents a self-calibrating line laser scanning system, which enables the creation of dense 3D models with a single fixed camera and a freely moving hand-held cross line laser projector. The proposed approach exploits geometric constraints, such as coplanarities, to recover the depth information and is applicable without any prior knowledge of the position and orientation of the laser projector. By employing an off-the-shelf underwater camera and a waterproof housing with high power line lasers an affordable 3D scanning solution can be built. In experiments the performance of the proposed technique is studied and compared with 3D reconstruction using explicit calibration. We demonstrate that the scanning system can be applied to above-the-water as well as underwater scenes.",
"We consider the problem of deliberately manipulating the direct and indirect light flowing through a time-varying, general scene in order to simplify its visual analysis. Our approach rests on a crucial link between stereo geometry and light transport: while direct light always obeys the epipolar geometry of a projector-camera pair, indirect light overwhelmingly does not. We show that it is possible to turn this observation into an imaging method that analyzes light transport in real time in the optical domain, prior to acquisition. This yields three key abilities that we demonstrate in an experimental camera prototype: (1) producing a live indirect-only video stream for any scene, regardless of geometric or photometric complexity; (2) capturing images that make existing structured-light shape recovery algorithms robust to indirect transport; and (3) turning them into one-shot methods for dynamic 3D shape capture.",
"",
"Underwater 3D shape scanning technique becomes popular because ofseveral rising research topics, such as map making ofsubmarine topography for autonomous underwater vehicle (UAV), shape measurement of live fish, motion capture of swimming human, etc. Structured light systems (SLS) based active 3D scanning systems are widely used in the air and also promising to apply underwater environment. When SLS is used in the air, the stereo correspondences can be efficiently retrieved by epipolar constraint. However, in the underwater environment, the camera and projector are usually set in special housings and refraction occurs at the interfaces between water glass and glass air, resulting in invalid conditions for epipolar constraint which severely deteriorates the correspondence search process. In this paper, we propose an efficient technique to calibrate the underwater SLS systems as well as robust 3D shape acquisition technique. In order to avoid the calculation complexity, we approximate the system with central projection model. Although such an approximation produces an inevitable errors in the system, such errors are diminished by a combination of grid based SLS technique and a bundle adjustment algorithm. We tested our method with a real underwater SLS, consisting ofcustom-made laser pattern projector and underwater housings, showing the validity ofour method.",
"A Laser-based Structured Light System (LbSLS) has been designed to perform underwater close-range 3D reconstructions even with high turbidity conditions and outperform conventional systems. The system uses a camera and a 532 nm green laser projector. The optical technique used is based on the projection of a pattern obtained placing a Diffractive Optical Element (DOE) in front of the laser beam. In the experiments described in this manuscript, the DOE used diffracts the laser beam in 25 parallel lines providing enough information in a single camera frame to perform a 3D reconstruction.",
"Commercial RGB-D cameras provide the possibility of fast, accurate, and cost-effective 3-D scanning solution in a single package. These economical depth cameras provide several advantages over conventional depth sensors, such as sonars and lidars, in specific usage scenarios. In this paper, we analyze the performance of Kinect v2 time-of-flight camera while operating fully submerged underwater in a customized waterproof housing. Camera calibration has been performed for Kinect’s RGB and NIR cameras, and the effect of calibration on the generated 3-D mesh is discussed in detail. To overcome the effect of refraction of light due to the sensor housing and water, we propose a time-of-flight correction method and a fast, accurate and intuitive refraction correction method that can be applied to the acquired depth images, during 3-D mesh generation. Experimental results show that the Kinect v2 can acquire point cloud data up to 650 mm. The reconstruction results have been analyzed qualitatively and quantitatively, and confirm that the 3-D reconstruction of submerged objects at small distances is possible without the requirement of any external NIR light source. The proposed algorithms successfully generated 3-D mesh with a mean error of ±6 mm at a frame rate of nearly 10 fps. We acquired a large data set of RGB, IR and depth data from a submerged Kinect v2. The data set covers a large variety of objects scanned underwater and is publicly available for further use, along with the Kinect waterproof housing design and correction filter codes. The research is aimed toward small-scale research activities and economical solution for 3-D scanning underwater. Applications such as coral reef mapping and underwater SLAM in shallow waters for ROV’s can be a viable application area that can benefit from results achieved.",
"We propose a new method to analyze light transport in homogeneous scattering media. The incident light undergoes multiple bounces in translucent objects, and produces a complex light field. Our method analyzes the light transport in two steps. First, single and multiple scattering are separated by projecting high-frequency stripe patterns. Then, multiple scattering is decomposed into each bounce component based on the light transport equation. The light field for each bounce is recursively estimated. Experimental results show that light transport in scattering media can be decomposed and visualized for each bounce."
]
}
|
1811.09675
|
2951583262
|
Dense 3D shape acquisition of swimming human or live fish is an important research topic for sports, biological science and so on. For this purpose, active stereo sensor is usually used in the air, however it cannot be applied to the underwater environment because of refraction, strong light attenuation and severe interference of bubbles. Passive stereo is a simple solution for capturing dynamic scenes at underwater environment, however the shape with textureless surfaces or irregular reflections cannot be recovered. Recently, the stereo camera pair with a pattern projector for adding artificial textures on the objects is proposed. However, to use the system for underwater environment, several problems should be compensated, i.e., disturbance by fluctuation and bubbles. Simple solution is to use convolutional neural network for stereo to cancel the effects of bubbles and or water fluctuation. Since it is not easy to train CNN with small size of database with large variation, we develop a special bubble generation device to efficiently create real bubble database of multiple size and density. In addition, we propose a transfer learning technique for multi-scale CNN to effectively remove bubbles and projected-patterns on the object. Further, we develop a real system and actually captured live swimming human, which has not been done before. Experiments are conducted to show the effectiveness of our method compared with the state of the art techniques.
|
Collection of huge data for learning is another open problem for CNN-based stereo techniques. For solution, Zhou al proposed a technique without using ground truth depth data, but LR consistency as a loss function @cite_11 . Tonioni al proposed a unsupervised method by using existing stereo technique as an instruction @cite_0 . Tulyakov and Ivanov proposed a multi-instance learning (MIL) method by using several constraints and cost functions @cite_35 . However in general, unsupervised learning is instable compared to supervised learning. DispNet @cite_23 and PSMNet @cite_5 are trained with generated images based on computer graphics, but transfer learning with natural images is necessary since computer graphics is not realistic enough to learn noises or camera characteristics. In this research, we created original stereo dataset and a special device for data augmentation which reproduces underwater environment for transfer learning.
|
{
"cite_N": [
"@cite_35",
"@cite_0",
"@cite_23",
"@cite_5",
"@cite_11"
],
"mid": [
"2559827556",
"2779124836",
"",
"2794812000",
"2776033207"
],
"abstract": [
"Deep-learning metrics have recently demonstrated extremely good performance to match image patches for stereo reconstruction. However, training such metrics requires large amount of labeled stereo images, which can be difficult or costly to collect for certain applications (consider, for example, satellite stereo imaging). The main contribution of our work is a new weakly supervised method for learning deep metrics from unlabeled stereo images, given coarse information about the scenes and the optical system. Our method alternatively optimizes the metric with a standard stochastic gradient descent, and applies stereo constraints to regularize its prediction. Experiments on reference data-sets show that, for a given network architecture, training with this new method without ground-truth produces a metric with performance as good as state-of-the-art baselines trained with the said ground-truth. This work has three practical implications. Firstly, it helps to overcome limitations of training sets, in particular noisy ground truth. Secondly it allows to use much more training data during learning. Thirdly, it allows to tune deep metric for a particular stereo system, even if ground truth is not available.",
"Recent ground-breaking works have shown that deep neural networks can be trained end-to-end to regress dense disparity maps directly from image pairs. Computer generated imagery is deployed to gather the large data corpus required to train such networks, an additional fine-tuning allowing to adapt the model to work well also on real and possibly diverse environments. Yet, besides a few public datasets such as Kitti, the ground-truth needed to adapt the network to a new scenario is hardly available in practice. In this paper we propose a novel unsupervised adaptation approach that enables to fine-tune a deep learning stereo model without any ground-truth information. We rely on off-the-shelf stereo algorithms together with state-of-the-art confidence measures, the latter able to ascertain upon correctness of the measurements yielded by former. Thus, we train the network based on a novel loss-function that penalizes predictions disagreeing with the highly confident disparities provided by the algorithm and enforces a smoothness constraint. Experiments on popular datasets (KITTI 2012, KITTI 2015 and Middlebury 2014) and other challenging test images demonstrate the effectiveness of our proposal.",
"",
"Recent work has shown that depth estimation from a stereo pair of images can be formulated as a supervised learning task to be resolved with convolutional neural networks (CNNs). However, current architectures rely on patch-based Siamese networks, lacking the means to exploit context information for finding correspondence in illposed regions. To tackle this problem, we propose PSMNet, a pyramid stereo matching network consisting of two main modules: spatial pyramid pooling and 3D CNN. The spatial pyramid pooling module takes advantage of the capacity of global context information by aggregating context in different scales and locations to form a cost volume. The 3D CNN learns to regularize cost volume using stacked multiple hourglass networks in conjunction with intermediate supervision. The proposed approach was evaluated on several benchmark datasets. Our method ranked first in the KITTI 2012 and 2015 leaderboards before March 18, 2018. The codes of PSMNet are available at: this https URL",
"Convolutional neural networks showed the ability in stereo matching cost learning. Recent approaches learned parameters from public datasets that have ground truth disparity maps. Due to the difficulty of labeling ground truth depth, usable data for system training is rather limited, making it difficult to apply the system to real applications. In this paper, we present a framework for learning stereo matching costs without human supervision. Our method updates network parameters in an iterative manner. It starts with a randomly initialized network. Left-right check is adopted to guide the training. Suitable matching is then picked and used as training data in following iterations. Our system finally converges to a stable state and performs even comparably with other supervised methods."
]
}
|
1811.09720
|
2949605285
|
We propose to explain the predictions of a deep neural network, by pointing to the set of what we call representer points in the training set, for a given test point prediction. Specifically, we show that we can decompose the pre-activation prediction of a neural network into a linear combination of activations of training points, with the weights corresponding to what we call representer values, which thus capture the importance of that training point on the learned parameters of the network. But it provides a deeper understanding of the network than simply training point influence: with positive representer values corresponding to excitatory training points, and negative values corresponding to inhibitory points, which as we show provides considerably more insight. Our method is also much more scalable, allowing for real-time feedback in a manner not feasible with influence functions.
|
Our approach is based on a representer theorem for deep neural network predictions. Representer theorems @cite_3 in machine learning contexts have focused on non-parametric regression, specifically in reproducing kernel Hilbert spaces (RKHS), and which loosely state that under certain conditions the minimizer of a loss functional over a RKHS can be expressed as a linear combination of kernel evaluations at training points. There have been recent efforts at leveraging such insights to compositional contexts , though these largely focus on connections to non-parametric estimation. extend the representer theorem to compositions of kernels, while draws connections between deep neural networks to such deep kernel estimation, specifically deep spline estimation. In our work, we consider the much simpler problem of explaining pre-activation neural network predictions in terms of activations of training points, which while less illuminating from a non-parametric estimation standpoint, is arguably much more explanatory, and useful from an explainable ML standpoint.
|
{
"cite_N": [
"@cite_3"
],
"mid": [
"1540155273"
],
"abstract": [
"Wahba's classical representer theorem states that the solutions of certain risk minimization problems involving an empirical risk term and a quadratic regularizer can be written as expansions in terms of the training examples. We generalize the theorem to a larger class of regularizers and empirical risk terms, and give a self-contained proof utilizing the feature space associated with a kernel. The result shows that a wide range of problems have optimal solutions that live in the finite dimensional span of the training examples mapped into feature space, thus enabling us to carry out kernel algorithms independent of the (potentially infinite) dimensionality of the feature space."
]
}
|
1811.09751
|
2900997043
|
When labeled data is scarce for a specific target task, transfer learning often offers an effective solution by utilizing data from a related source task. However, when transferring knowledge from a less related source, it may inversely hurt the target performance, a phenomenon known as negative transfer. Despite its pervasiveness, negative transfer is usually described in an informal manner, lacking rigorous definition, careful analysis, or systematic treatment. This paper proposes a formal definition of negative transfer and analyzes three important aspects thereof. Stemming from this analysis, a novel technique is proposed to circumvent negative transfer by filtering out unrelated source data. Based on adversarial networks, the technique is highly generic and can be applied to a wide range of transfer learning algorithms. The proposed approach is evaluated on six state-of-the-art deep transfer methods via experiments on four benchmark datasets with varying levels of difficulty. Empirically, the proposed method consistently improves the performance of all baseline methods and largely avoids negative transfer, even when the source data is degenerate.
|
@cite_1 @cite_4 uses knowledge learned in the source domain to assist training in the target domain. Early methods exploit conventional statistical techniques such as instance weighting @cite_28 and feature mapping @cite_34 @cite_38 . Compared to these earlier approaches, deep transfer networks achieve better results in discovering domain invariant factors @cite_6 . Some deep methods @cite_7 @cite_36 transfer via distribution (mis)match measurements such as Maximum Mean Discrepancy (MMD) @cite_28 . More recent work @cite_26 @cite_29 @cite_5 @cite_11 exploit generative adversarial networks (GANs) @cite_18 and add a subnetwork as a domain discriminator. These methods achieve state-of-the-art on computer vision tasks @cite_11 and some natural language processing tasks @cite_12 . However, none of these techniques are specifically designed to tackle the problem of negative transfer.
|
{
"cite_N": [
"@cite_38",
"@cite_18",
"@cite_26",
"@cite_4",
"@cite_7",
"@cite_28",
"@cite_36",
"@cite_29",
"@cite_1",
"@cite_6",
"@cite_5",
"@cite_34",
"@cite_12",
"@cite_11"
],
"mid": [
"143806433",
"",
"",
"1992946551",
"2159291411",
"2811380766",
"2467286621",
"2214409633",
"2165698076",
"2149933564",
"2738463471",
"2115403315",
"2740100794",
"2605488490"
],
"abstract": [
"Common assumption in most machine learning algorithms is that, labeled (source) data and unlabeled (target) data are sampled from the same distribution. However, many real world tasks violate this assumption: in temporal domains, feature distributions may vary over time, clinical studies may have sampling bias, or sometimes sufficient labeled data for the domain of interest does not exist, and labeled data from a related domain must be utilized. In such settings, knowing in which dimensions source and target data vary is extremely important to reduce the distance between domains and accurately transfer knowledge. In this paper, we present a novel method to identify variant and invariant features between two datasets. Our contribution is two fold: First, we present a novel transfer learning approach for domain adaptation, and second, we formalize the problem of finding differently distributed features as a convex optimization problem. Experimental studies on synthetic and benchmark real world datasets show that our approach outperform other transfer learning approaches, and it aids the prediction accuracy significantly.",
"",
"",
"We explore a transfer learning setting, in which a finite sequence of target concepts are sampled independently with an unknown distribution from a known family. We study the total number of labeled examples required to learn all targets to an arbitrary specified expected accuracy, focusing on the asymptotics in the number of tasks and the desired accuracy. Our primary interest is formally understanding the fundamental benefits of transfer learning, compared to learning each target independently from the others. Our approach to the transfer problem is general, in the sense that it can be used with a variety of learning protocols. As a particularly interesting application, we study in detail the benefits of transfer for self-verifying active learning; in this setting, we find that the number of labeled examples required for learning with transfer is often significantly smaller than that required for learning each target independently.",
"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 consider the scenario where training and test data are drawn from different distributions, commonly referred to as sample selection bias. Most algorithms for this setting try to first recover sampling distributions and then make appropriate corrections based on the distribution estimate. We present a nonparametric method which directly produces resampling weights without distribution estimation. Our method works by matching distributions between training and testing sets in feature space. Experimental results demonstrate that our method works well in practice.",
"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 is a \"frustratingly easy\" 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.",
"Recent reports suggest that a generic supervised deep CNN model trained on a large-scale dataset reduces, but does not remove, dataset bias. Fine-tuning deep models in a new domain can require a significant amount of labeled data, which for many applications is simply not available. We propose a new CNN architecture to exploit unlabeled and sparsely labeled target domain data. Our approach simultaneously optimizes for domain invariance to facilitate domain transfer and uses a soft label distribution matching loss to transfer information between tasks. Our proposed adaptation method offers empirical performance which exceeds previously published results on two standard benchmark visual domain adaptation tasks, evaluated across supervised and semi-supervised adaptation settings.",
"A major assumption in many machine learning and data mining algorithms is that the training and future data must be in the same feature space and have the same distribution. However, in many real-world applications, this assumption may not hold. For example, we sometimes have a classification task in one domain of interest, but we only have sufficient training data in another domain of interest, where the latter data may be in a different feature space or follow a different data distribution. In such cases, knowledge transfer, if done successfully, would greatly improve the performance of learning by avoiding much expensive data-labeling efforts. In recent years, transfer learning has emerged as a new learning framework to address this problem. This survey focuses on categorizing and reviewing the current progress on transfer learning for classification, regression, and clustering problems. In this survey, we discuss the relationship between transfer learning and other related machine learning techniques such as domain adaptation, multitask learning and sample selection bias, as well as covariate shift. We also explore some potential future issues in transfer learning research.",
"Many deep neural networks trained on natural images exhibit a curious phenomenon in common: on the first layer they learn features similar to Gabor filters and color blobs. Such first-layer features appear not to be specific to a particular dataset or task, but general in that they are applicable to many datasets and tasks. Features must eventually transition from general to specific by the last layer of the network, but this transition has not been studied extensively. In this paper we experimentally quantify the generality versus specificity of neurons in each layer of a deep convolutional neural network and report a few surprising results. Transferability is negatively affected by two distinct issues: (1) the specialization of higher layer neurons to their original task at the expense of performance on the target task, which was expected, and (2) optimization difficulties related to splitting networks between co-adapted neurons, which was not expected. In an example network trained on ImageNet, we demonstrate that either of these two issues may dominate, depending on whether features are transferred from the bottom, middle, or top of the network. We also document that the transferability of features decreases as the distance between the base task and target task increases, but that transferring features even from distant tasks can be better than using random features. A final surprising result is that initializing a network with transferred features from almost any number of layers can produce a boost to generalization that lingers even after fine-tuning to the target dataset.",
"Adversarial learning has been successfully embedded into deep networks to learn transferable features, which reduce distribution discrepancy between the source and target domains. Existing domain adversarial networks assume fully shared label space across domains. In the presence of big data, there is strong motivation of transferring both classification and representation models from existing big domains to unknown small domains. This paper introduces partial transfer learning, which relaxes the shared label space assumption to that the target label space is only a subspace of the source label space. Previous methods typically match the whole source domain to the target domain, which are prone to negative transfer for the partial transfer problem. We present Selective Adversarial Network (SAN), which simultaneously circumvents negative transfer by selecting out the outlier source classes and promotes positive transfer by maximally matching the data distributions in the shared label space. Experiments demonstrate that our models exceed state-of-the-art results for partial transfer learning tasks on several benchmark datasets.",
"Domain adaptation allows knowledge from a source domain to be transferred to a different but related target domain. Intuitively, discovering a good feature representation across domains is crucial. In this paper, we first propose to find such a representation through a new learning method, transfer component analysis (TCA), for domain adaptation. TCA tries to learn some transfer components across domains in a reproducing kernel Hilbert space using maximum mean miscrepancy. In the subspace spanned by these transfer components, data properties are preserved and data distributions in different domains are close to each other. As a result, with the new representations in this subspace, we can apply standard machine learning methods to train classifiers or regression models in the source domain for use in the target domain. Furthermore, in order to uncover the knowledge hidden in the relations between the data labels from the source and target domains, we extend TCA in a semisupervised learning setting, which encodes label information into transfer components learning. We call this extension semisupervised TCA. The main contribution of our work is that we propose a novel dimensionality reduction framework for reducing the distance between domains in a latent space for domain adaptation. We propose both unsupervised and semisupervised feature extraction approaches, which can dramatically reduce the distance between domain distributions by projecting data onto the learned transfer components. Finally, our approach can handle large datasets and naturally lead to out-of-sample generalization. The effectiveness and efficiency of our approach are verified by experiments on five toy datasets and two real-world applications: cross-domain indoor WiFi localization and cross-domain text classification.",
"",
"Domain Adaptation is an actively researched problem in Computer Vision. In this work, we propose an approach that leverages unsupervised data to bring the source and target distributions closer in a learned joint feature space. We accomplish this by inducing a symbiotic relationship between the learned embedding and a generative adversarial network. This is in contrast to methods which use the adversarial framework for realistic data generation and retraining deep models with such data. We demonstrate the strength and generality of our approach by performing experiments on three different tasks with varying levels of difficulty: (1) Digit classification (MNIST, SVHN and USPS datasets) (2) Object recognition using OFFICE dataset and (3) Domain adaptation from synthetic to real data. Our method achieves state-of-the art performance in most experimental settings and by far the only GAN-based method that has been shown to work well across different datasets such as OFFICE and DIGITS."
]
}
|
1811.09751
|
2900997043
|
When labeled data is scarce for a specific target task, transfer learning often offers an effective solution by utilizing data from a related source task. However, when transferring knowledge from a less related source, it may inversely hurt the target performance, a phenomenon known as negative transfer. Despite its pervasiveness, negative transfer is usually described in an informal manner, lacking rigorous definition, careful analysis, or systematic treatment. This paper proposes a formal definition of negative transfer and analyzes three important aspects thereof. Stemming from this analysis, a novel technique is proposed to circumvent negative transfer by filtering out unrelated source data. Based on adversarial networks, the technique is highly generic and can be applied to a wide range of transfer learning algorithms. The proposed approach is evaluated on six state-of-the-art deep transfer methods via experiments on four benchmark datasets with varying levels of difficulty. Empirically, the proposed method consistently improves the performance of all baseline methods and largely avoids negative transfer, even when the source data is degenerate.
|
Early work that noted negative transfer @cite_33 was targeted at simple classifiers such as hierarchical Naive Bayes. Later, similar negative effects have also been observed in various settings including multi-source transfer learning @cite_16 , imbalanced distributions @cite_3 and partial transfer learning @cite_5 . While the importance of detecting and avoiding negative transfer has raised increasing attention @cite_20 , the literature lacks in-depth analysis.
|
{
"cite_N": [
"@cite_33",
"@cite_3",
"@cite_5",
"@cite_16",
"@cite_20"
],
"mid": [
"2835011589",
"1919803322",
"2738463471",
"2122084318",
"2395579298"
],
"abstract": [
"Multi-source transfer learning has been proven effective when within-target labeled data is scarce. Previous work focuses primarily on exploiting domain similarities and assumes that source domains are richly or at least comparably labeled. While this strong assumption is never true in practice, this paper relaxes it and addresses challenges related to sources with diverse labeling volume and diverse reliability. The first challenge is combining domain similarity and source reliability by proposing a new transfer learning method that utilizes both source-target similarities and inter-source relationships. The second challenge involves pool-based active learning where the oracle is only available in source domains, resulting in an integrated active transfer learning framework that incorporates distribution matching and uncertainty sampling. Extensive experiments on synthetic and two real-world datasets clearly demonstrate the superiority of our proposed methods over several baselines including state-of-the-art transfer learning methods. Code related to this paper is available at: https: github.com iedwardwangi ReliableMSTL.",
"Transfer learning has benefited many real-world applications where labeled data are abundant in source domains but scarce in the target domain. As there are usually multiple relevant domains where knowledge can be transferred, multiple source transfer learning MSTL has recently attracted much attention. However, we are facing two major challenges when applying MSTL. First, without knowledge about the difference between source and target domains, negative transfer occurs when knowledge is transferred from highly irrelevant sources. Second, existence of imbalanced distributions in classes, where examples in one class dominate, can lead to improper judgement on the source domains' relevance to the target task. Since existing MSTL methods are usually designed to transfer from relevant sources with balanced distributions, they will fail in applications where these two challenges persist. In this article, we propose a novel two-phase framework to effectively transfer knowledge from multiple sources even when there exists irrelevant sources and imbalanced class distributions. First, an effective supervised local weight scheme is proposed to assign a proper weight to each source domain's classifier based on its ability of predicting accurately on each local region of the target domain. The second phase then learns a classifier for the target domain by solving an optimization problem which concerns both training error minimization and consistency with weighted predictions gained from source domains. A theoretical analysis shows that as the number of source domains increases, the probability that the proposed approach has an error greater than a bound is becoming exponentially small. We further extend the proposed approach to an online processing scenario to conduct transfer learning on continuously arriving data. Extensive experiments on disease prediction, spam filtering and intrusion detection datasets demonstrate that: i the proposed two-phase approach outperforms existing MSTL approaches due to its ability of tackling negative transfer and imbalanced distribution challenges, and ii the proposed online approach achieves comparable performance to the offline scheme.",
"Adversarial learning has been successfully embedded into deep networks to learn transferable features, which reduce distribution discrepancy between the source and target domains. Existing domain adversarial networks assume fully shared label space across domains. In the presence of big data, there is strong motivation of transferring both classification and representation models from existing big domains to unknown small domains. This paper introduces partial transfer learning, which relaxes the shared label space assumption to that the target label space is only a subspace of the source label space. Previous methods typically match the whole source domain to the target domain, which are prone to negative transfer for the partial transfer problem. We present Selective Adversarial Network (SAN), which simultaneously circumvents negative transfer by selecting out the outlier source classes and promotes positive transfer by maximally matching the data distributions in the shared label space. Experiments demonstrate that our models exceed state-of-the-art results for partial transfer learning tasks on several benchmark datasets.",
"Recent work has demonstrated the effectiveness of domain adaptation methods for computer vision applications. In this work, we propose a new multiple source domain adaptation method called Domain Selection Machine (DSM) for event recognition in consumer videos by leveraging a large number of loosely labeled web images from different sources (e.g., Flickr.com and Photosig.com), in which there are no labeled consumer videos. Specifically, we first train a set of SVM classifiers (referred to as source classifiers) by using the SIFT features of web images from different source domains. We propose a new parametric target decision function to effectively integrate the static SIFT features from web images video keyframes and the spacetime (ST) features from consumer videos. In order to select the most relevant source domains, we further introduce a new data-dependent regularizer into the objective of Support Vector Regression (SVR) using the ∊-insensitive loss, which enforces the target classifier shares similar decision values on the unlabeled consumer videos with the selected source classifiers. Moreover, we develop an alternating optimization algorithm to iteratively solve the target decision function and a domain selection vector which indicates the most relevant source domains. Extensive experiments on three real-world datasets demonstrate the effectiveness of our proposed method DSM over the state-of-the-art by a performance gain up to 46.41 .",
"Machine learning and data mining techniques have been used in numerous real-world applications. An assumption of traditional machine learning methodologies is the training data and testing data are taken from the same domain, such that the input feature space and data distribution characteristics are the same. However, in some real-world machine learning scenarios, this assumption does not hold. There are cases where training data is expensive or difficult to collect. Therefore, there is a need to create high-performance learners trained with more easily obtained data from different domains. This methodology is referred to as transfer learning. This survey paper formally defines transfer learning, presents information on current solutions, and reviews applications applied to transfer learning. Lastly, there is information listed on software downloads for various transfer learning solutions and a discussion of possible future research work. The transfer learning solutions surveyed are independent of data size and can be applied to big data environments."
]
}
|
1811.09953
|
2901469102
|
Homomorphic encryption enables arbitrary computation over data while it remains encrypted. This privacy-preserving feature is attractive for machine learning, but requires significant computational time due to the large overhead of the encryption scheme. We present Faster CryptoNets, a method for efficient encrypted inference using neural networks. We develop a pruning and quantization approach that leverages sparse representations in the underlying cryptosystem to accelerate inference. We derive an optimal approximation for popular activation functions that achieves maximally-sparse encodings and minimizes approximation error. We also show how privacy-safe training techniques can be used to reduce the overhead of encrypted inference for real-world datasets by leveraging transfer learning and differential privacy. Our experiments show that our method maintains competitive accuracy and achieves a significant speedup over previous methods. This work increases the viability of deep learning systems that use homomorphic encryption to protect user privacy.
|
Differential privacy allows statistics to be computed over a dataset without revealing information about individual records @cite_3 @cite_52 . A common method is to apply noise to individual examples to obfuscate statistical differences that might be distinguishable @cite_50 . However, differential privacy is better suited for the training phase. During test-time, adding noise to a single example may change the prediction.
|
{
"cite_N": [
"@cite_50",
"@cite_52",
"@cite_3"
],
"mid": [
"2950602864",
"",
"2109426455"
],
"abstract": [
"Some machine learning applications involve training data that is sensitive, such as the medical histories of patients in a clinical trial. A model may inadvertently and implicitly store some of its training data; careful analysis of the model may therefore reveal sensitive information. To address this problem, we demonstrate a generally applicable approach to providing strong privacy guarantees for training data: Private Aggregation of Teacher Ensembles (PATE). The approach combines, in a black-box fashion, multiple models trained with disjoint datasets, such as records from different subsets of users. Because they rely directly on sensitive data, these models are not published, but instead used as \"teachers\" for a \"student\" model. The student learns to predict an output chosen by noisy voting among all of the teachers, and cannot directly access an individual teacher or the underlying data or parameters. The student's privacy properties can be understood both intuitively (since no single teacher and thus no single dataset dictates the student's training) and formally, in terms of differential privacy. These properties hold even if an adversary can not only query the student but also inspect its internal workings. Compared with previous work, the approach imposes only weak assumptions on how teachers are trained: it applies to any model, including non-convex models like DNNs. We achieve state-of-the-art privacy utility trade-offs on MNIST and SVHN thanks to an improved privacy analysis and semi-supervised learning.",
"",
"Over the past five years a new approach to privacy-preserving data analysis has born fruit [13, 18, 7, 19, 5, 37, 35, 8, 32]. This approach differs from much (but not all!) of the related literature in the statistics, databases, theory, and cryptography communities, in that a formal and ad omnia privacy guarantee is defined, and the data analysis techniques presented are rigorously proved to satisfy the guarantee. The key privacy guarantee that has emerged is differential privacy. Roughly speaking, this ensures that (almost, and quantifiably) no risk is incurred by joining a statistical database. In this survey, we recall the definition of differential privacy and two basic techniques for achieving it. We then show some interesting applications of these techniques, presenting algorithms for three specific tasks and three general results on differentially private learning."
]
}
|
1811.09953
|
2901469102
|
Homomorphic encryption enables arbitrary computation over data while it remains encrypted. This privacy-preserving feature is attractive for machine learning, but requires significant computational time due to the large overhead of the encryption scheme. We present Faster CryptoNets, a method for efficient encrypted inference using neural networks. We develop a pruning and quantization approach that leverages sparse representations in the underlying cryptosystem to accelerate inference. We derive an optimal approximation for popular activation functions that achieves maximally-sparse encodings and minimizes approximation error. We also show how privacy-safe training techniques can be used to reduce the overhead of encrypted inference for real-world datasets by leveraging transfer learning and differential privacy. Our experiments show that our method maintains competitive accuracy and achieves a significant speedup over previous methods. This work increases the viability of deep learning systems that use homomorphic encryption to protect user privacy.
|
Secure multi-party computation enables multiple parties to jointly compute a function over their inputs while keeping their inputs private. This has been explored using Garbled Circuits @cite_68 in the works of @cite_55 @cite_51 and @cite_39 . These methods often involve a high communication complexity with significant bandwidth costs. Fully homomorphic encryption (FHE) was proposed by @cite_54 and allows anyone to compute over encrypted data without decrypting it @cite_36 . A weaker version of FHE, termed leveled homomorphic encryption (LHE) permits a subset of arithmetic operations on a depth-bounded arithmetic circuit @cite_70 . While HE has been explored for machine learning applications, many works focus on simpler models such as linear @cite_66 , logistic @cite_37 and ridge regression @cite_48 . CryptoNets @cite_57 was one of the first works to implement HE in a neural network setting. More recently, @cite_22 and @cite_4 extended this to deeper network architectures and developed additional polynomial approximations to the activation function that leveraged batch normalization for stability.
|
{
"cite_N": [
"@cite_37",
"@cite_4",
"@cite_22",
"@cite_36",
"@cite_70",
"@cite_54",
"@cite_55",
"@cite_48",
"@cite_39",
"@cite_57",
"@cite_68",
"@cite_51",
"@cite_66"
],
"mid": [
"2770953739",
"",
"",
"",
"1992282993",
"2031533839",
"",
"",
"",
"2435473771",
"2088492763",
"",
"162878246"
],
"abstract": [
"In 2014, introduced a cloud service scenario to provide private predictive analyses on encrypted medical data, and gave a proof of concept implementation by utilizing homomorphic encryption (HE) scheme. In their implementation, they needed to approximate an analytic predictive model to a polynomial, using Taylor approximations. However, their approach could not reach a satisfactory compromise so that they just restricted the pool of data to guarantee suitable accuracy. In this paper, we suggest and implement a new efficient approach to provide the service using minimax approximation and Non-Adjacent Form (NAF) encoding. With our method, it is possible to remove the limitation of input range and reduce maximum errors, allowing faster analyses than the previous work. Moreover, we prove that the NAF encoding allows us to use more efficient parameters than the binary encoding used in the previous work or balaced base-B encoding. For comparison with the previous work, we present implementation results using HElib. Our implementation gives a prediction with 7-bit precision (of maximal error 0.0044) for having a heart attack, and makes the prediction in 0.5 s on a single laptop. We also implement the private healthcare service analyzing a Cox Proportional Hazard Model for the first time.",
"",
"",
"",
"We present a fully homomorphic encryption scheme that is based solely on the(standard) learning with errors (LWE) assumption. Applying known results on LWE, the security of our scheme is based on the worst-case hardness of short vector problems'' on arbitrary lattices. Our construction improves on previous works in two aspects: We show that somewhat homomorphic'' encryption can be based on LWE, using a new re-linearization technique. In contrast, all previous schemes relied on complexity assumptions related to ideals in various rings. We deviate from the \"squashing paradigm'' used in all previous works. We introduce a new dimension-modulus reduction technique, which shortens the cipher texts and reduces the decryption complexity of our scheme, without introducing additional assumptions . Our scheme has very short cipher texts and we therefore use it to construct an asymptotically efficient LWE-based single-server private information retrieval (PIR) protocol. The communication complexity of our protocol (in the public-key model) is @math bits per single-bit query (here, @math is a security parameter).",
"We propose a fully homomorphic encryption scheme -- i.e., a scheme that allows one to evaluate circuits over encrypted data without being able to decrypt. Our solution comes in three steps. First, we provide a general result -- that, to construct an encryption scheme that permits evaluation of arbitrary circuits, it suffices to construct an encryption scheme that can evaluate (slightly augmented versions of) its own decryption circuit; we call a scheme that can evaluate its (augmented) decryption circuit bootstrappable. Next, we describe a public key encryption scheme using ideal lattices that is almost bootstrappable. Lattice-based cryptosystems typically have decryption algorithms with low circuit complexity, often dominated by an inner product computation that is in NC1. Also, ideal lattices provide both additive and multiplicative homomorphisms (modulo a public-key ideal in a polynomial ring that is represented as a lattice), as needed to evaluate general circuits. Unfortunately, our initial scheme is not quite bootstrappable -- i.e., the depth that the scheme can correctly evaluate can be logarithmic in the lattice dimension, just like the depth of the decryption circuit, but the latter is greater than the former. In the final step, we show how to modify the scheme to reduce the depth of the decryption circuit, and thereby obtain a bootstrappable encryption scheme, without reducing the depth that the scheme can evaluate. Abstractly, we accomplish this by enabling the encrypter to start the decryption process, leaving less work for the decrypter, much like the server leaves less work for the decrypter in a server-aided cryptosystem.",
"",
"",
"",
"Applying machine learning to a problem which involves medical, financial, or other types of sensitive data, not only requires accurate predictions but also careful attention to maintaining data privacy and security. Legal and ethical requirements may prevent the use of cloud-based machine learning solutions for such tasks. In this work, we will present a method to convert learned neural networks to CryptoNets, neural networks that can be applied to encrypted data. This allows a data owner to send their data in an encrypted form to a cloud service that hosts the network. The encryption ensures that the data remains confidential since the cloud does not have access to the keys needed to decrypt it. Nevertheless, we will show that the cloud service is capable of applying the neural network to the encrypted data to make encrypted predictions, and also return them in encrypted form. These encrypted predictions can be sent back to the owner of the secret key who can decrypt them. Therefore, the cloud service does not gain any information about the raw data nor about the prediction it made. We demonstrate CryptoNets on the MNIST optical character recognition tasks. CryptoNets achieve 99 accuracy and can make around 59000 predictions per hour on a single PC. Therefore, they allow high throughput, accurate, and private predictions.",
"In this paper we introduce a new tool for controlling the knowledge transfer process in cryptographic protocol design. It is applied to solve a general class of problems which include most of the two-party cryptographic problems in the literature. Specifically, we show how two parties A and B can interactively generate a random integer N = p?q such that its secret, i.e., the prime factors (p, q), is hidden from either party individually but is recoverable jointly if desired. This can be utilized to give a protocol for two parties with private values i and j to compute any polynomially computable functions f(i,j) and g(i,j) with minimal knowledge transfer and a strong fairness property. As a special case, A and B can exchange a pair of secrets sA, sB, e.g. the factorization of an integer and a Hamiltonian circuit in a graph, in such a way that sA becomes computable by B when and only when sB becomes computable by A. All these results are proved assuming only that the problem of factoring large intergers is computationally intractable.",
"",
""
]
}
|
1811.09763
|
2901161195
|
The research on hashing techniques for visual data is gaining increased attention in recent years due to the need for compact representations supporting efficient search retrieval in large-scale databases such as online images. Among many possibilities, Mean Average Precision(mAP) has emerged as the dominant performance metric for hashing-based retrieval. One glaring shortcoming of mAP is its inability in balancing retrieval accuracy and utilization of hash codes: pushing a system to attain higher mAP will inevitably lead to poorer utilization of the hash codes. Poor utilization of the hash codes hinders good retrieval because of increased collision of samples in the hash space. This means that a model giving a higher mAP values does not necessarily do a better job in retrieval. In this paper, we introduce a new metric named Mean Local Group Average Precision (mLGAP) for better evaluation of the performance of hashing-based retrieval. The new metric provides a retrieval performance measure that also reconciles the utilization of hash codes, leading to a more practically meaningful performance metric than conventional ones like mAP. To this end, we start by mathematical analysis of the deficiencies of mAP for hashing-based retrieval. We then propose mLGAP and show why it is more appropriate for hashing-based retrieval. Experiments on image retrieval are used to demonstrate the effectiveness of the proposed metric.
|
Hashing techniques can be divided into two categories: data-independent methods, and data-dependent methods. One representative data-independent approach is Locality-Sensitive Hashing (LSH) @cite_15 , which uses random projections to generate hash functions. LSH has been extended to several versions, such as Kernelized LSH @cite_16 and other variants @cite_17 @cite_20 . However, empirical results suggest that, LSH and other data-dependent approaches like @cite_13 usually require long bit length to maintain high precision and recall. This not only lowers the speed performance, but also increases space complexity. Hence data-dependent approaches are not deemed as the best option for large-scale problems like Internet-scale image retrieval.
|
{
"cite_N": [
"@cite_15",
"@cite_16",
"@cite_13",
"@cite_20",
"@cite_17"
],
"mid": [
"1502916507",
"2171790913",
"2144892774",
"",
"2162006472"
],
"abstract": [
"The nearestor near-neighbor query problems arise in a large variety of database applications, usually in the context of similarity searching. Of late, there has been increasing interest in building search index structures for performing similarity search over high-dimensional data, e.g., image databases, document collections, time-series databases, and genome databases. Unfortunately, all known techniques for solving this problem fall prey to the of dimensionality.\" That is, the data structures scale poorly with data dimensionality; in fact, if the number of dimensions exceeds 10 to 20, searching in k-d trees and related structures involves the inspection of a large fraction of the database, thereby doing no better than brute-force linear search. It has been suggested that since the selection of features and the choice of a distance metric in typical applications is rather heuristic, determining an approximate nearest neighbor should su ce for most practical purposes. In this paper, we examine a novel scheme for approximate similarity search based on hashing. The basic idea is to hash the points Supported by NAVY N00014-96-1-1221 grant and NSF Grant IIS-9811904. Supported by Stanford Graduate Fellowship and NSF NYI Award CCR-9357849. Supported by ARO MURI Grant DAAH04-96-1-0007, NSF Grant IIS-9811904, and NSF Young Investigator Award CCR9357849, with matching funds from IBM, Mitsubishi, Schlumberger Foundation, Shell Foundation, and Xerox Corporation. Permission to copy without fee all or part of this material is granted provided that the copies are not made or distributed for direct commercial advantage, the VLDB copyright notice and the title of the publication and its date appear, and notice is given that copying is by permission of the Very Large Data Base Endowment. To copy otherwise, or to republish, requires a fee and or special permission from the Endowment. Proceedings of the 25th VLDB Conference, Edinburgh, Scotland, 1999. from the database so as to ensure that the probability of collision is much higher for objects that are close to each other than for those that are far apart. We provide experimental evidence that our method gives signi cant improvement in running time over other methods for searching in highdimensional spaces based on hierarchical tree decomposition. Experimental results also indicate that our scheme scales well even for a relatively large number of dimensions (more than 50).",
"Fast retrieval methods are critical for large-scale and data-driven vision applications. Recent work has explored ways to embed high-dimensional features or complex distance functions into a low-dimensional Hamming space where items can be efficiently searched. However, existing methods do not apply for high-dimensional kernelized data when the underlying feature embedding for the kernel is unknown. We show how to generalize locality-sensitive hashing to accommodate arbitrary kernel functions, making it possible to preserve the algorithm's sub-linear time similarity search guarantees for a wide class of useful similarity functions. Since a number of successful image-based kernels have unknown or incomputable embeddings, this is especially valuable for image retrieval tasks. We validate our technique on several large-scale datasets, and show that it enables accurate and fast performance for example-based object classification, feature matching, and content-based retrieval.",
"We introduce a method that enables scalable similarity search for learned metrics. Given pairwise similarity and dissimilarity constraints between some examples, we learn a Mahalanobis distance function that captures the examples' underlying relationships well. To allow sublinear time similarity search under the learned metric, we show how to encode the learned metric parameterization into randomized locality-sensitive hash functions. We further formulate an indirect solution that enables metric learning and hashing for vector spaces whose high dimensionality makes it infeasible to learn an explicit transformation over the feature dimensions. We demonstrate the approach applied to a variety of image data sets, as well as a systems data set. The learned metrics improve accuracy relative to commonly used metric baselines, while our hashing construction enables efficient indexing with learned distances and very large databases.",
"",
"We present a novel Locality-Sensitive Hashing scheme for the Approximate Nearest Neighbor Problem under lp norm, based on p-stable distributions.Our scheme improves the running time of the earlier algorithm for the case of the lp norm. It also yields the first known provably efficient approximate NN algorithm for the case p<1. We also show that the algorithm finds the exact near neigbhor in O(log n) time for data satisfying certain \"bounded growth\" condition.Unlike earlier schemes, our LSH scheme works directly on points in the Euclidean space without embeddings. Consequently, the resulting query time bound is free of large factors and is simple and easy to implement. Our experiments (on synthetic data sets) show that the our data structure is up to 40 times faster than kd-tree."
]
}
|
1811.09763
|
2901161195
|
The research on hashing techniques for visual data is gaining increased attention in recent years due to the need for compact representations supporting efficient search retrieval in large-scale databases such as online images. Among many possibilities, Mean Average Precision(mAP) has emerged as the dominant performance metric for hashing-based retrieval. One glaring shortcoming of mAP is its inability in balancing retrieval accuracy and utilization of hash codes: pushing a system to attain higher mAP will inevitably lead to poorer utilization of the hash codes. Poor utilization of the hash codes hinders good retrieval because of increased collision of samples in the hash space. This means that a model giving a higher mAP values does not necessarily do a better job in retrieval. In this paper, we introduce a new metric named Mean Local Group Average Precision (mLGAP) for better evaluation of the performance of hashing-based retrieval. The new metric provides a retrieval performance measure that also reconciles the utilization of hash codes, leading to a more practically meaningful performance metric than conventional ones like mAP. To this end, we start by mathematical analysis of the deficiencies of mAP for hashing-based retrieval. We then propose mLGAP and show why it is more appropriate for hashing-based retrieval. Experiments on image retrieval are used to demonstrate the effectiveness of the proposed metric.
|
On the other hand, data-dependent approaches are supposed to generate shorter hash codes, since more data-specific information can be exploited. Since the space spanned by meaningful images is in general only a small portion of the entire vector space, by using a machine-learning strategy, it is possible to tailor a hashing scheme to cater to this space so that more compact binary codes may be obtained ( the images can be represented by shorter binary hash codes). Several hashing techniques in this category, like @cite_19 @cite_3 @cite_12 @cite_4 , have been proposed, reporting promising performance. These techniques can be further divided into unsupervised approaches and supervised approaches. Unsupervised approaches use unlabeled training data to learn the hash functions. Representative algorithms include PCA hasing @cite_27 , which is based on principal component analysis (PCA); Iterative Quantization (ITQ) @cite_1 , which applies orthogonal rotation matrices to tune the initial projection matrix learned by PCA; and Spectral Hashing (SH) @cite_0 , which is based on the eigenvectors computed from the data similarity graph. Deep Hashing (DH) @cite_2 is another example, which leverages the capacity of neural networks on acquiring better visual features for improved performance.
|
{
"cite_N": [
"@cite_4",
"@cite_1",
"@cite_3",
"@cite_0",
"@cite_19",
"@cite_27",
"@cite_2",
"@cite_12"
],
"mid": [
"2142881874",
"2084363474",
"1992371516",
"",
"2293824885",
"2074668987",
"1956333070",
"2221852422"
],
"abstract": [
"Hashing has emerged as a popular technique for fast nearest neighbor search in gigantic databases. In particular, learning based hashing has received considerable attention due to its appealing storage and search efficiency. However, the performance of most unsupervised learning based hashing methods deteriorates rapidly as the hash code length increases. We argue that the degraded performance is due to inferior optimization procedures used to achieve discrete binary codes. This paper presents a graph-based unsupervised hashing model to preserve the neighborhood structure of massive data in a discrete code space. We cast the graph hashing problem into a discrete optimization framework which directly learns the binary codes. A tractable alternating maximization algorithm is then proposed to explicitly deal with the discrete constraints, yielding high-quality codes to well capture the local neighborhoods. Extensive experiments performed on four large datasets with up to one million samples show that our discrete optimization based graph hashing method obtains superior search accuracy over state-of-the-art un-supervised hashing methods, especially for longer codes.",
"This paper addresses the problem of learning similarity-preserving binary codes for efficient retrieval in large-scale image collections. We propose a simple and efficient alternating minimization scheme for finding a rotation of zero-centered data so as to minimize the quantization error of mapping this data to the vertices of a zero-centered binary hypercube. This method, dubbed iterative quantization (ITQ), has connections to multi-class spectral clustering and to the orthogonal Procrustes problem, and it can be used both with unsupervised data embeddings such as PCA and supervised embeddings such as canonical correlation analysis (CCA). Our experiments show that the resulting binary coding schemes decisively outperform several other state-of-the-art methods.",
"Recent years have witnessed the growing popularity of hashing in large-scale vision problems. It has been shown that the hashing quality could be boosted by leveraging supervised information into hash function learning. However, the existing supervised methods either lack adequate performance or often incur cumbersome model training. In this paper, we propose a novel kernel-based supervised hashing model which requires a limited amount of supervised information, i.e., similar and dissimilar data pairs, and a feasible training cost in achieving high quality hashing. The idea is to map the data to compact binary codes whose Hamming distances are minimized on similar pairs and simultaneously maximized on dissimilar pairs. Our approach is distinct from prior works by utilizing the equivalence between optimizing the code inner products and the Hamming distances. This enables us to sequentially and efficiently train the hash functions one bit at a time, yielding very short yet discriminative codes. We carry out extensive experiments on two image benchmarks with up to one million samples, demonstrating that our approach significantly outperforms the state-of-the-arts in searching both metric distance neighbors and semantically similar neighbors, with accuracy gains ranging from 13 to 46 .",
"",
"Hashing is a popular approximate nearest neighbor search approach for large-scale image retrieval. Supervised hashing, which incorporates similarity dissimilarity information on entity pairs to improve the quality of hashing function learning, has recently received increasing attention. However, in the existing supervised hashing methods for images, an input image is usually encoded by a vector of handcrafted visual features. Such hand-crafted feature vectors do not necessarily preserve the accurate semantic similarities of images pairs, which may often degrade the performance of hashing function learning. In this paper, we propose a supervised hashing method for image retrieval, in which we automatically learn a good image representation tailored to hashing as well as a set of hash functions. The proposed method has two stages. In the first stage, given the pairwise similarity matrix S over training images, we propose a scalable coordinate descent method to decompose S into a product of HHT where H is a matrix with each of its rows being the approximate hash code associated to a training image. In the second stage, we propose to simultaneously learn a good feature representation for the input images as well as a set of hash functions, via a deep convolutional network tailored to the learned hash codes in H and optionally the discrete class labels of the images. Extensive empirical evaluations on three benchmark datasets with different kinds of images show that the proposed method has superior performance gains over several state-of-the-art supervised and unsupervised hashing methods.",
"Hashing-based approximate nearest neighbor (ANN) search in huge databases has become popular due to its computational and memory efficiency. The popular hashing methods, e.g., Locality Sensitive Hashing and Spectral Hashing, construct hash functions based on random or principal projections. The resulting hashes are either not very accurate or are inefficient. Moreover, these methods are designed for a given metric similarity. On the contrary, semantic similarity is usually given in terms of pairwise labels of samples. There exist supervised hashing methods that can handle such semantic similarity, but they are prone to overfitting when labeled data are small or noisy. In this work, we propose a semi-supervised hashing (SSH) framework that minimizes empirical error over the labeled set and an information theoretic regularizer over both labeled and unlabeled sets. Based on this framework, we present three different semi-supervised hashing methods, including orthogonal hashing, nonorthogonal hashing, and sequential hashing. Particularly, the sequential hashing method generates robust codes in which each hash function is designed to correct the errors made by the previous ones. We further show that the sequential learning paradigm can be extended to unsupervised domains where no labeled pairs are available. Extensive experiments on four large datasets (up to 80 million samples) demonstrate the superior performance of the proposed SSH methods over state-of-the-art supervised and unsupervised hashing techniques.",
"In this paper, we propose a new deep hashing (DH) approach to learn compact binary codes for large scale visual search. Unlike most existing binary codes learning methods which seek a single linear projection to map each sample into a binary vector, we develop a deep neural network to seek multiple hierarchical non-linear transformations to learn these binary codes, so that the nonlinear relationship of samples can be well exploited. Our model is learned under three constraints at the top layer of the deep network: 1) the loss between the original real-valued feature descriptor and the learned binary vector is minimized, 2) the binary codes distribute evenly on each bit, and 3) different bits are as independent as possible. To further improve the discriminative power of the learned binary codes, we extend DH into supervised DH (SDH) by including one discriminative term into the objective function of DH which simultaneously maximizes the inter-class variations and minimizes the intra-class variations of the learned binary codes. Experimental results show the superiority of the proposed approach over the state-of-the-arts.",
"We propose a method for learning similarity-preserving hash functions that map high-dimensional data onto binary codes. The formulation is based on structured prediction with latent variables and a hinge-like loss function. It is efficient to train for large datasets, scales well to large code lengths, and outperforms state-of-the-art methods."
]
}
|
1811.09763
|
2901161195
|
The research on hashing techniques for visual data is gaining increased attention in recent years due to the need for compact representations supporting efficient search retrieval in large-scale databases such as online images. Among many possibilities, Mean Average Precision(mAP) has emerged as the dominant performance metric for hashing-based retrieval. One glaring shortcoming of mAP is its inability in balancing retrieval accuracy and utilization of hash codes: pushing a system to attain higher mAP will inevitably lead to poorer utilization of the hash codes. Poor utilization of the hash codes hinders good retrieval because of increased collision of samples in the hash space. This means that a model giving a higher mAP values does not necessarily do a better job in retrieval. In this paper, we introduce a new metric named Mean Local Group Average Precision (mLGAP) for better evaluation of the performance of hashing-based retrieval. The new metric provides a retrieval performance measure that also reconciles the utilization of hash codes, leading to a more practically meaningful performance metric than conventional ones like mAP. To this end, we start by mathematical analysis of the deficiencies of mAP for hashing-based retrieval. We then propose mLGAP and show why it is more appropriate for hashing-based retrieval. Experiments on image retrieval are used to demonstrate the effectiveness of the proposed metric.
|
With training data that come with label information, such as point-wise labels, pairwise labels @cite_19 @cite_25 , or triplet labels @cite_14 , supervised methods have been developed to take advantage of the extra information. Well-known supervised approaches include Supervised Hashing with Kernels (KSH) @cite_3 , which learns the hash function in a kernel space; Minimal Loss Hashing (MLH) @cite_12 , which minimizes a hinge loss function to learn the hash function; Binary Reconstructive Embeddings (BRE) @cite_7 , which learns hash functions by minimizing the reconstruction error between the vectors from the original space and the Hamming space; and Supervised Deep Hashing (SDH) @cite_2 , which learns the binary codes by a deep neural network. While these methods use relaxation schemes to obtain the discrete binary codes, Discrete Graph Hashing (DGH) @cite_4 and Supervised Discrete Hashing (SDiscH) @cite_26 were also proposed to calculate the optimal binary codes directly, and improved performance was reported.
|
{
"cite_N": [
"@cite_14",
"@cite_4",
"@cite_26",
"@cite_7",
"@cite_3",
"@cite_19",
"@cite_2",
"@cite_25",
"@cite_12"
],
"mid": [
"1939575207",
"2142881874",
"1910300841",
"2164338181",
"1992371516",
"2293824885",
"1956333070",
"",
"2221852422"
],
"abstract": [
"Similarity-preserving hashing is a widely-used method for nearest neighbour search in large-scale image retrieval tasks. For most existing hashing methods, an image is first encoded as a vector of hand-engineering visual features, followed by another separate projection or quantization step that generates binary codes. However, such visual feature vectors may not be optimally compatible with the coding process, thus producing sub-optimal hashing codes. In this paper, we propose a deep architecture for supervised hashing, in which images are mapped into binary codes via carefully designed deep neural networks. The pipeline of the proposed deep architecture consists of three building blocks: 1) a sub-network with a stack of convolution layers to produce the effective intermediate image features; 2) a divide-and-encode module to divide the intermediate image features into multiple branches, each encoded into one hash bit; and 3) a triplet ranking loss designed to characterize that one image is more similar to the second image than to the third one. Extensive evaluations on several benchmark image datasets show that the proposed simultaneous feature learning and hash coding pipeline brings substantial improvements over other state-of-the-art supervised or unsupervised hashing methods.",
"Hashing has emerged as a popular technique for fast nearest neighbor search in gigantic databases. In particular, learning based hashing has received considerable attention due to its appealing storage and search efficiency. However, the performance of most unsupervised learning based hashing methods deteriorates rapidly as the hash code length increases. We argue that the degraded performance is due to inferior optimization procedures used to achieve discrete binary codes. This paper presents a graph-based unsupervised hashing model to preserve the neighborhood structure of massive data in a discrete code space. We cast the graph hashing problem into a discrete optimization framework which directly learns the binary codes. A tractable alternating maximization algorithm is then proposed to explicitly deal with the discrete constraints, yielding high-quality codes to well capture the local neighborhoods. Extensive experiments performed on four large datasets with up to one million samples show that our discrete optimization based graph hashing method obtains superior search accuracy over state-of-the-art un-supervised hashing methods, especially for longer codes.",
"Recently, learning based hashing techniques have attracted broad research interests because they can support efficient storage and retrieval for high-dimensional data such as images, videos, documents, etc. However, a major difficulty of learning to hash lies in handling the discrete constraints imposed on the pursued hash codes, which typically makes hash optimizations very challenging (NP-hard in general). In this work, we propose a new supervised hashing framework, where the learning objective is to generate the optimal binary hash codes for linear classification. By introducing an auxiliary variable, we reformulate the objective such that it can be solved substantially efficiently by employing a regularization algorithm. One of the key steps in this algorithm is to solve a regularization sub-problem associated with the NP-hard binary optimization. We show that the sub-problem admits an analytical solution via cyclic coordinate descent. As such, a high-quality discrete solution can eventually be obtained in an efficient computing manner, therefore enabling to tackle massive datasets. We evaluate the proposed approach, dubbed Supervised Discrete Hashing (SDH), on four large image datasets and demonstrate its superiority to the state-of-the-art hashing methods in large-scale image retrieval.",
"Fast retrieval methods are increasingly critical for many large-scale analysis tasks, and there have been several recent methods that attempt to learn hash functions for fast and accurate nearest neighbor searches. In this paper, we develop an algorithm for learning hash functions based on explicitly minimizing the reconstruction error between the original distances and the Hamming distances of the corresponding binary embeddings. We develop a scalable coordinate-descent algorithm for our proposed hashing objective that is able to efficiently learn hash functions in a variety of settings. Unlike existing methods such as semantic hashing and spectral hashing, our method is easily kernelized and does not require restrictive assumptions about the underlying distribution of the data. We present results over several domains to demonstrate that our method outperforms existing state-of-the-art techniques.",
"Recent years have witnessed the growing popularity of hashing in large-scale vision problems. It has been shown that the hashing quality could be boosted by leveraging supervised information into hash function learning. However, the existing supervised methods either lack adequate performance or often incur cumbersome model training. In this paper, we propose a novel kernel-based supervised hashing model which requires a limited amount of supervised information, i.e., similar and dissimilar data pairs, and a feasible training cost in achieving high quality hashing. The idea is to map the data to compact binary codes whose Hamming distances are minimized on similar pairs and simultaneously maximized on dissimilar pairs. Our approach is distinct from prior works by utilizing the equivalence between optimizing the code inner products and the Hamming distances. This enables us to sequentially and efficiently train the hash functions one bit at a time, yielding very short yet discriminative codes. We carry out extensive experiments on two image benchmarks with up to one million samples, demonstrating that our approach significantly outperforms the state-of-the-arts in searching both metric distance neighbors and semantically similar neighbors, with accuracy gains ranging from 13 to 46 .",
"Hashing is a popular approximate nearest neighbor search approach for large-scale image retrieval. Supervised hashing, which incorporates similarity dissimilarity information on entity pairs to improve the quality of hashing function learning, has recently received increasing attention. However, in the existing supervised hashing methods for images, an input image is usually encoded by a vector of handcrafted visual features. Such hand-crafted feature vectors do not necessarily preserve the accurate semantic similarities of images pairs, which may often degrade the performance of hashing function learning. In this paper, we propose a supervised hashing method for image retrieval, in which we automatically learn a good image representation tailored to hashing as well as a set of hash functions. The proposed method has two stages. In the first stage, given the pairwise similarity matrix S over training images, we propose a scalable coordinate descent method to decompose S into a product of HHT where H is a matrix with each of its rows being the approximate hash code associated to a training image. In the second stage, we propose to simultaneously learn a good feature representation for the input images as well as a set of hash functions, via a deep convolutional network tailored to the learned hash codes in H and optionally the discrete class labels of the images. Extensive empirical evaluations on three benchmark datasets with different kinds of images show that the proposed method has superior performance gains over several state-of-the-art supervised and unsupervised hashing methods.",
"In this paper, we propose a new deep hashing (DH) approach to learn compact binary codes for large scale visual search. Unlike most existing binary codes learning methods which seek a single linear projection to map each sample into a binary vector, we develop a deep neural network to seek multiple hierarchical non-linear transformations to learn these binary codes, so that the nonlinear relationship of samples can be well exploited. Our model is learned under three constraints at the top layer of the deep network: 1) the loss between the original real-valued feature descriptor and the learned binary vector is minimized, 2) the binary codes distribute evenly on each bit, and 3) different bits are as independent as possible. To further improve the discriminative power of the learned binary codes, we extend DH into supervised DH (SDH) by including one discriminative term into the objective function of DH which simultaneously maximizes the inter-class variations and minimizes the intra-class variations of the learned binary codes. Experimental results show the superiority of the proposed approach over the state-of-the-arts.",
"",
"We propose a method for learning similarity-preserving hash functions that map high-dimensional data onto binary codes. The formulation is based on structured prediction with latent variables and a hinge-like loss function. It is efficient to train for large datasets, scales well to large code lengths, and outperforms state-of-the-art methods."
]
}
|
1811.09729
|
2901865741
|
Detecting manipulated images has become a significant emerging challenge. The advent of image sharing platforms and the easy availability of advanced photo editing software have resulted in a large quantities of manipulated images being shared on the internet. While the intent behind such manipulations varies widely, concerns on the spread of fake news and misinformation is growing. Current state of the art methods for detecting these manipulated images suffers from the lack of training data due to the laborious labeling process. We address this problem in this paper, for which we introduce a manipulated image generation process that creates true positives using currently available datasets. Drawing from traditional work on image blending, we propose a novel generator for creating such examples. In addition, we also propose to further create examples that force the algorithm to focus on boundary artifacts during training. Strong experimental results validate our proposal.
|
. GAN based image editing approaches have witnessed a rapid emergence and impressive results recently @cite_12 @cite_18 @cite_38 @cite_29 . Prior and concurrent works force the output of GAN to be conditioned on input images through extra regression losses ( ., @math loss) or discrete labels. In particular, Tsai al. @cite_12 generate natural composite images using both scene parsing and harmonized ground truth. Pathak al. @cite_38 present a context encoder trained with reconstruction plus an adversarial loss to inpaint missing image contents. In contrast to these methods, we generate manipulated images for better generalization ability of a manipulation segmentation network.
|
{
"cite_N": [
"@cite_38",
"@cite_18",
"@cite_29",
"@cite_12"
],
"mid": [
"2342877626",
"2164147879",
"2963800363",
"2962737447"
],
"abstract": [
"We present an unsupervised visual feature learning algorithm driven by context-based pixel prediction. By analogy with auto-encoders, we propose Context Encoders -- a convolutional neural network trained to generate the contents of an arbitrary image region conditioned on its surroundings. In order to succeed at this task, context encoders need to both understand the content of the entire image, as well as produce a plausible hypothesis for the missing part(s). When training context encoders, we have experimented with both a standard pixel-wise reconstruction loss, as well as a reconstruction plus an adversarial loss. The latter produces much sharper results because it can better handle multiple modes in the output. We found that a context encoder learns a representation that captures not just appearance but also the semantics of visual structures. We quantitatively demonstrate the effectiveness of our learned features for CNN pre-training on classification, detection, and segmentation tasks. Furthermore, context encoders can be used for semantic inpainting tasks, either stand-alone or as initialization for non-parametric methods.",
"Why does placing an object from one photograph into another often make the colors of that object suddenly look wrong? One possibility is that humans prefer distributions of colors that are often found in nature; that is, we find pleasing these color combinations that we see often. Another possibility is that humans simply prefer colors to be consistent within an image, regardless of what they are. In this paper, we explore some of these issues by studying the color statistics of a large dataset of natural images, and by looking at differences in color distribution in realistic and unrealistic images. We apply our findings to two problems: 1) classifying composite images into realistic vs. non- realistic, and 2) recoloring image regions for realistic compositing.",
"We present a new method for synthesizing high-resolution photo-realistic images from semantic label maps using conditional generative adversarial networks (conditional GANs). Conditional GANs have enabled a variety of applications, but the results are often limited to low-resolution and still far from realistic. In this work, we generate 2048 A— 1024 visually appealing results with a novel adversarial loss, as well as new multi-scale generator and discriminator architectures. Furthermore, we extend our framework to interactive visual manipulation with two additional features. First, we incorporate object instance segmentation information, which enables object manipulations such as removing adding objects and changing the object category. Second, we propose a method to generate diverse results given the same input, allowing users to edit the object appearance interactively. Human opinion studies demonstrate that our method significantly outperforms existing methods, advancing both the quality and the resolution of deep image synthesis and editing.",
"Compositing is one of the most common operations in photo editing. To generate realistic composites, the appearances of foreground and background need to be adjusted to make them compatible. Previous approaches to harmonize composites have focused on learning statistical relationships between hand-crafted appearance features of the foreground and background, which is unreliable especially when the contents in the two layers are vastly different. In this work, we propose an end-to-end deep convolutional neural network for image harmonization, which can capture both the context and semantic information of the composite images during harmonization. We also introduce an efficient way to collect large-scale and high-quality training data that can facilitate the training process. Experiments on the synthesized dataset and real composite images show that the proposed network outperforms previous state-of-the-art methods."
]
}
|
1811.09729
|
2901865741
|
Detecting manipulated images has become a significant emerging challenge. The advent of image sharing platforms and the easy availability of advanced photo editing software have resulted in a large quantities of manipulated images being shared on the internet. While the intent behind such manipulations varies widely, concerns on the spread of fake news and misinformation is growing. Current state of the art methods for detecting these manipulated images suffers from the lack of training data due to the laborious labeling process. We address this problem in this paper, for which we introduce a manipulated image generation process that creates true positives using currently available datasets. Drawing from traditional work on image blending, we propose a novel generator for creating such examples. In addition, we also propose to further create examples that force the algorithm to focus on boundary artifacts during training. Strong experimental results validate our proposal.
|
. Discriminative feature learning has motivated recent research on adversarial training on several tasks. Shrivastava al. @cite_0 propose a simulated and unsupervised learning approach which utilizes synthetic images to generate realistic images. Wang al. @cite_22 boost the performance on occluded and deformed objects through an online hard negative generation network. Wei al. @cite_44 investigate an adversarial erasing approach to learn dense and complete semantic segmentation. Le al. @cite_27 propose an adversarial shadow attenuation network to make correct predictions on hard shadow examples. Inspired by these works and taking into account the demand for diverse examples in the image manipulation segmentation task, we generate both hard and easy examples to help the network learn manipulation artifacts.
|
{
"cite_N": [
"@cite_0",
"@cite_44",
"@cite_27",
"@cite_22"
],
"mid": [
"2963709863",
"2600144439",
"2884217841",
"2607037079"
],
"abstract": [
"With recent progress in graphics, it has become more tractable to train models on synthetic images, potentially avoiding the need for expensive annotations. However, learning from synthetic images may not achieve the desired performance due to a gap between synthetic and real image distributions. To reduce this gap, we propose Simulated+Unsupervised (S+U) learning, where the task is to learn a model to improve the realism of a simulators output using unlabeled real data, while preserving the annotation information from the simulator. We develop a method for S+U learning that uses an adversarial network similar to Generative Adversarial Networks (GANs), but with synthetic images as inputs instead of random vectors. We make several key modifications to the standard GAN algorithm to preserve annotations, avoid artifacts, and stabilize training: (i) a self-regularization term, (ii) a local adversarial loss, and (iii) updating the discriminator using a history of refined images. We show that this enables generation of highly realistic images, which we demonstrate both qualitatively and with a user study. We quantitatively evaluate the generated images by training models for gaze estimation and hand pose estimation. We show a significant improvement over using synthetic images, and achieve state-of-the-art results on the MPIIGaze dataset without any labeled real data.",
"We investigate a principle way to progressively mine discriminative object regions using classification networks to address the weakly-supervised semantic segmentation problems. Classification networks are only responsive to small and sparse discriminative regions from the object of interest, which deviates from the requirement of the segmentation task that needs to localize dense, interior and integral regions for pixel-wise inference. To mitigate this gap, we propose a new adversarial erasing approach for localizing and expanding object regions progressively. Starting with a single small object region, our proposed approach drives the classification network to sequentially discover new and complement object regions by erasing the current mined regions in an adversarial manner. These localized regions eventually constitute a dense and complete object region for learning semantic segmentation. To further enhance the quality of the discovered regions by adversarial erasing, an online prohibitive segmentation learning approach is developed to collaborate with adversarial erasing by providing auxiliary segmentation supervision modulated by the more reliable classification scores. Despite its apparent simplicity, the proposed approach achieves 55.0 and 55.7 mean Intersection-over-Union (mIoU) scores on PASCAL VOC 2012 val and test sets, which are the new state-of-the-arts.",
"We propose a novel GAN-based framework for detecting shadows in images, in which a shadow detection network (D-Net) is trained together with a shadow attenuation network (A-Net) that generates adversarial training examples. The A-Net modifies the original training images constrained by a simplified physical shadow model and is focused on fooling the D-Net’s shadow predictions. Hence, it is effectively augmenting the training data for D-Net with hard-to-predict cases. The D-Net is trained to predict shadows in both original images and generated images from the A-Net. Our experimental results show that the additional training data from A-Net significantly improves the shadow detection accuracy of D-Net. Our method outperforms the state-of-the-art methods on the most challenging shadow detection benchmark (SBU) and also obtains state-of-the-art results on a cross-dataset task, testing on UCF. Furthermore, the proposed method achieves accurate real-time shadow detection at 45 frames per second.",
"How do we learn an object detector that is invariant to occlusions and deformations? Our current solution is to use a data-driven strategy – collect large-scale datasets which have object instances under different conditions. The hope is that the final classifier can use these examples to learn invariances. But is it really possible to see all the occlusions in a dataset? We argue that like categories, occlusions and object deformations also follow a long-tail. Some occlusions and deformations are so rare that they hardly happen, yet we want to learn a model invariant to such occurrences. In this paper, we propose an alternative solution. We propose to learn an adversarial network that generates examples with occlusions and deformations. The goal of the adversary is to generate examples that are difficult for the object detector to classify. In our framework both the original detector and adversary are learned in a joint manner. Our experimental results indicate a 2.3 mAP boost on VOC07 and a 2.6 mAP boost on VOC2012 object detection challenge compared to the Fast-RCNN pipeline."
]
}
|
1811.09789
|
2901974839
|
There has been much recent work on image captioning models that describe the factual aspects of an image. Recently, some models have incorporated non-factual aspects into the captions, such as sentiment or style. However, such models typically have difficulty in balancing the semantic aspects of the image and the non-factual dimensions of the caption; in addition, it can be observed that humans may focus on different aspects of an image depending on the chosen sentiment or style of the caption. To address this, we design an attention-based model to better add sentiment to image captions. The model embeds and learns sentiment with respect to image-caption data, and uses both high-level and word-level sentiment information during the learning process. The model outperforms the state-of-the-art work in image captioning with sentiment using standard evaluation metrics. An analysis of generated captions also shows that our model does this by a better selection of the sentiment-bearing adjectives and adjective-noun pairs.
|
Image captioning systems are usually designed according to a top-down paradigm and include a combination of a Convolutional Neural Network (CNN) and a long short-term memory (LSTM) to encode the visual content and generate the image descriptions, respectively @cite_21 , which is inspired from the work of Sutskever al @cite_9 . The current state-of-the-art models in image captioning are attention-based systems @cite_19 @cite_10 @cite_16 @cite_24 . The models use visual content, referred to as spatial features, as the input of an attention mechanism to selectively attend to different parts of an image at each time step in generating the image caption.
|
{
"cite_N": [
"@cite_9",
"@cite_21",
"@cite_24",
"@cite_19",
"@cite_16",
"@cite_10"
],
"mid": [
"2949888546",
"2951912364",
"2950178297",
"2951590222",
"2963084599",
"2952469094"
],
"abstract": [
"Deep Neural Networks (DNNs) are powerful models that have achieved excellent performance on difficult learning tasks. Although DNNs work well whenever large labeled training sets are available, they cannot be used to map sequences to sequences. In this paper, we present a general end-to-end approach to sequence learning that makes minimal assumptions on the sequence structure. Our method uses a multilayered Long Short-Term Memory (LSTM) to map the input sequence to a vector of a fixed dimensionality, and then another deep LSTM to decode the target sequence from the vector. Our main result is that on an English to French translation task from the WMT'14 dataset, the translations produced by the LSTM achieve a BLEU score of 34.8 on the entire test set, where the LSTM's BLEU score was penalized on out-of-vocabulary words. Additionally, the LSTM did not have difficulty on long sentences. For comparison, a phrase-based SMT system achieves a BLEU score of 33.3 on the same dataset. When we used the LSTM to rerank the 1000 hypotheses produced by the aforementioned SMT system, its BLEU score increases to 36.5, which is close to the previous best result on this task. The LSTM also learned sensible phrase and sentence representations that are sensitive to word order and are relatively invariant to the active and the passive voice. Finally, we found that reversing the order of the words in all source sentences (but not target sentences) improved the LSTM's performance markedly, because doing so introduced many short term dependencies between the source and the target sentence which made the optimization problem easier.",
"Automatically describing the content of an image is a fundamental problem in artificial intelligence that connects computer vision and natural language processing. In this paper, we present a generative model based on a deep recurrent architecture that combines recent advances in computer vision and machine translation and that can be used to generate natural sentences describing an image. The model is trained to maximize the likelihood of the target description sentence given the training image. Experiments on several datasets show the accuracy of the model and the fluency of the language it learns solely from image descriptions. Our model is often quite accurate, which we verify both qualitatively and quantitatively. For instance, while the current state-of-the-art BLEU-1 score (the higher the better) on the Pascal dataset is 25, our approach yields 59, to be compared to human performance around 69. We also show BLEU-1 score improvements on Flickr30k, from 56 to 66, and on SBU, from 19 to 28. Lastly, on the newly released COCO dataset, we achieve a BLEU-4 of 27.7, which is the current state-of-the-art.",
"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.",
"Top-down visual attention mechanisms have been used extensively in image captioning and visual question answering (VQA) to enable deeper image understanding through fine-grained analysis and even multiple steps of reasoning. In this work, we propose a combined bottom-up and top-down attention mechanism that enables attention to be calculated at the level of objects and other salient image regions. This is the natural basis for attention to be considered. Within our approach, the bottom-up mechanism (based on Faster R-CNN) proposes image regions, each with an associated feature vector, while the top-down mechanism determines feature weightings. Applying this approach to image captioning, our results on the MSCOCO test server establish a new state-of-the-art for the task, achieving CIDEr SPICE BLEU-4 scores of 117.9, 21.5 and 36.9, respectively. Demonstrating the broad applicability of the method, applying the same approach to VQA we obtain first place in the 2017 VQA Challenge.",
"Recently it has been shown that policy-gradient methods for reinforcement learning can be utilized to train deep end-to-end systems directly on non-differentiable metrics for the task at hand. In this paper we consider the problem of optimizing image captioning systems using reinforcement learning, and show that by carefully optimizing our systems using the test metrics of the MSCOCO task, significant gains in performance can be realized. Our systems are built using a new optimization approach that we call self-critical sequence training (SCST). SCST is a form of the popular REINFORCE algorithm that, rather than estimating a baseline to normalize the rewards and reduce variance, utilizes the output of its own test-time inference algorithm to normalize the rewards it experiences. Using this approach, estimating the reward signal (as actor-critic methods must do) and estimating normalization (as REINFORCE algorithms typically do) is avoided, while at the same time harmonizing the model with respect to its test-time inference procedure. Empirically we find that directly optimizing the CIDEr metric with SCST and greedy decoding at test-time is highly effective. Our results on the MSCOCO evaluation sever establish a new state-of-the-art on the task, improving the best result in terms of CIDEr from 104.9 to 114.7.",
"Attention-based neural encoder-decoder frameworks have been widely adopted for image captioning. Most methods force visual attention to be active for every generated word. However, the decoder likely requires little to no visual information from the image to predict non-visual words such as \"the\" and \"of\". Other words that may seem visual can often be predicted reliably just from the language model e.g., \"sign\" after \"behind a red stop\" or \"phone\" following \"talking on a cell\". In this paper, we propose a novel adaptive attention model with a visual sentinel. At each time step, our model decides whether to attend to the image (and if so, to which regions) or to the visual sentinel. The model decides whether to attend to the image and where, in order to extract meaningful information for sequential word generation. We test our method on the COCO image captioning 2015 challenge dataset and Flickr30K. Our approach sets the new state-of-the-art by a significant margin."
]
}
|
1811.09789
|
2901974839
|
There has been much recent work on image captioning models that describe the factual aspects of an image. Recently, some models have incorporated non-factual aspects into the captions, such as sentiment or style. However, such models typically have difficulty in balancing the semantic aspects of the image and the non-factual dimensions of the caption; in addition, it can be observed that humans may focus on different aspects of an image depending on the chosen sentiment or style of the caption. To address this, we design an attention-based model to better add sentiment to image captions. The model embeds and learns sentiment with respect to image-caption data, and uses both high-level and word-level sentiment information during the learning process. The model outperforms the state-of-the-art work in image captioning with sentiment using standard evaluation metrics. An analysis of generated captions also shows that our model does this by a better selection of the sentiment-bearing adjectives and adjective-noun pairs.
|
Yu al @cite_25 and You al @cite_29 applied a notion of Semantic attention to detected visual attributes, which is learned in an end-to-end fashion. This attention model is used to attend to semantic concepts detected from various parts of a given image. Here, they used the visual content only in the initial time step. In other time steps, Semantic attention was used to select the extracted semantic concepts. That is, Semantic attention differs from spatial attention, which attends to spatial features in every time step, and does not preserve the spatial information of the detected concepts.
|
{
"cite_N": [
"@cite_29",
"@cite_25"
],
"mid": [
"2953022248",
"2949118724"
],
"abstract": [
"Automatically generating a natural language description of an image has attracted interests recently both because of its importance in practical applications and because it connects two major artificial intelligence fields: computer vision and natural language processing. Existing approaches are either top-down, which start from a gist of an image and convert it into words, or bottom-up, which come up with words describing various aspects of an image and then combine them. In this paper, we propose a new algorithm that combines both approaches through a model of semantic attention. Our algorithm learns to selectively attend to semantic concept proposals and fuse them into hidden states and outputs of recurrent neural networks. The selection and fusion form a feedback connecting the top-down and bottom-up computation. We evaluate our algorithm on two public benchmarks: Microsoft COCO and Flickr30K. Experimental results show that our algorithm significantly outperforms the state-of-the-art approaches consistently across different evaluation metrics.",
"We propose a high-level concept word detector that can be integrated with any video-to-language models. It takes a video as input and generates a list of concept words as useful semantic priors for language generation models. The proposed word detector has two important properties. First, it does not require any external knowledge sources for training. Second, the proposed word detector is trainable in an end-to-end manner jointly with any video-to-language models. To maximize the values of detected words, we also develop a semantic attention mechanism that selectively focuses on the detected concept words and fuse them with the word encoding and decoding in the language model. In order to demonstrate that the proposed approach indeed improves the performance of multiple video-to-language tasks, we participate in four tasks of LSMDC 2016. Our approach achieves the best accuracies in three of them, including fill-in-the-blank, multiple-choice test, and movie retrieval. We also attain comparable performance for the other task, movie description."
]
}
|
1811.09789
|
2901974839
|
There has been much recent work on image captioning models that describe the factual aspects of an image. Recently, some models have incorporated non-factual aspects into the captions, such as sentiment or style. However, such models typically have difficulty in balancing the semantic aspects of the image and the non-factual dimensions of the caption; in addition, it can be observed that humans may focus on different aspects of an image depending on the chosen sentiment or style of the caption. To address this, we design an attention-based model to better add sentiment to image captions. The model embeds and learns sentiment with respect to image-caption data, and uses both high-level and word-level sentiment information during the learning process. The model outperforms the state-of-the-art work in image captioning with sentiment using standard evaluation metrics. An analysis of generated captions also shows that our model does this by a better selection of the sentiment-bearing adjectives and adjective-noun pairs.
|
To preserve the spatial information, the salient regions are localized using spatial transformer networks @cite_22 , which get the spatial features as inputs. This is similar to Faster-RCNN generation of bounding boxes @cite_15 , but it is trained in an end-to-end fashion using bilinear interpolation instead of a Region of interest pooling mechanism @cite_3 . Similarly, Anderson al @cite_19 applied spatial features, but using a pre-trained Faster-RCNN and an attention mechanism to discriminate among different visual-based concepts regarding the spatial features.
|
{
"cite_N": [
"@cite_19",
"@cite_15",
"@cite_22",
"@cite_3"
],
"mid": [
"2951590222",
"639708223",
"2951005624",
"2963758027"
],
"abstract": [
"Top-down visual attention mechanisms have been used extensively in image captioning and visual question answering (VQA) to enable deeper image understanding through fine-grained analysis and even multiple steps of reasoning. In this work, we propose a combined bottom-up and top-down attention mechanism that enables attention to be calculated at the level of objects and other salient image regions. This is the natural basis for attention to be considered. Within our approach, the bottom-up mechanism (based on Faster R-CNN) proposes image regions, each with an associated feature vector, while the top-down mechanism determines feature weightings. Applying this approach to image captioning, our results on the MSCOCO test server establish a new state-of-the-art for the task, achieving CIDEr SPICE BLEU-4 scores of 117.9, 21.5 and 36.9, respectively. Demonstrating the broad applicability of the method, applying the same approach to VQA we obtain first place in the 2017 VQA Challenge.",
"State-of-the-art object detection networks depend on region proposal algorithms to hypothesize object locations. Advances like SPPnet [1] and Fast R-CNN [2] 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. The RPN is trained end-to-end to generate high-quality region proposals, which are used by Fast R-CNN for detection. We further merge RPN and Fast R-CNN into a single network by sharing their convolutional features—using the recently popular terminology of neural networks with ’attention’ mechanisms, the RPN component tells the unified network where to look. For the very deep VGG-16 model [3] , our detection system has a frame rate of 5 fps ( including all steps ) on a GPU, while achieving state-of-the-art object detection accuracy on PASCAL VOC 2007, 2012, and MS COCO datasets with only 300 proposals per image. In ILSVRC and COCO 2015 competitions, Faster R-CNN and RPN are the foundations of the 1st-place winning entries in several tracks. Code has been made publicly available.",
"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.",
"We introduce the dense captioning task, which requires a computer vision system to both localize and describe salient regions in images in natural language. The dense captioning task generalizes object detection when the descriptions consist of a single word, and Image Captioning when one predicted region covers the full image. To address the localization and description task jointly we propose a Fully Convolutional Localization Network (FCLN) architecture that processes an image with a single, efficient forward pass, requires no external regions proposals, and can be trained end-to-end with a single round of optimization. The architecture is composed of a Convolutional Network, a novel dense localization layer, and Recurrent Neural Network language model that generates the label sequences. We evaluate our network on the Visual Genome dataset, which comprises 94,000 images and 4,100,000 region-grounded captions. We observe both speed and accuracy improvements over baselines based on current state of the art approaches in both generation and retrieval settings."
]
}
|
1811.09789
|
2901974839
|
There has been much recent work on image captioning models that describe the factual aspects of an image. Recently, some models have incorporated non-factual aspects into the captions, such as sentiment or style. However, such models typically have difficulty in balancing the semantic aspects of the image and the non-factual dimensions of the caption; in addition, it can be observed that humans may focus on different aspects of an image depending on the chosen sentiment or style of the caption. To address this, we design an attention-based model to better add sentiment to image captions. The model embeds and learns sentiment with respect to image-caption data, and uses both high-level and word-level sentiment information during the learning process. The model outperforms the state-of-the-art work in image captioning with sentiment using standard evaluation metrics. An analysis of generated captions also shows that our model does this by a better selection of the sentiment-bearing adjectives and adjective-noun pairs.
|
Recently, Hu al @cite_18 used variational autoencoders to control a generated sentence in terms of its attributes including sentiment and tense: they conditioned the sentence encoding process on these attributes. In conversation generation, Zhou al @cite_4 used emotion categories to control the responses in terms of emotions. As a part of their system, they fed an embedded emotion as an input to their decoder. Ghosh al @cite_31 proposed a model conditioning conversational text generation using affect categories. The model can control a generated sentence without previous knowledge about the words in the vocabulary. In our work, in contrast, we feed in embedded sentiments to capture both high-level and word-level sentiment information.
|
{
"cite_N": [
"@cite_31",
"@cite_18",
"@cite_4"
],
"mid": [
"2949378066",
"2735642330",
"2605133118"
],
"abstract": [
"Human verbal communication includes affective messages which are conveyed through use of emotionally colored words. There has been a lot of research in this direction but the problem of integrating state-of-the-art neural language models with affective information remains an area ripe for exploration. In this paper, we propose an extension to an LSTM (Long Short-Term Memory) language model for generating conversational text, conditioned on affect categories. Our proposed model, Affect-LM enables us to customize the degree of emotional content in generated sentences through an additional design parameter. Perception studies conducted using Amazon Mechanical Turk show that Affect-LM generates naturally looking emotional sentences without sacrificing grammatical correctness. Affect-LM also learns affect-discriminative word representations, and perplexity experiments show that additional affective information in conversational text can improve language model prediction.",
"Generic generation and manipulation of text is challenging and has limited success compared to recent deep generative modeling in visual domain. This paper aims at generating plausible natural language sentences, whose attributes are dynamically controlled by learning disentangled latent representations with designated semantics. We propose a new neural generative model which combines variational auto-encoders and holistic attribute discriminators for effective imposition of semantic structures. With differentiable approximation to discrete text samples, explicit constraints on independent attribute controls, and efficient collaborative learning of generator and discriminators, our model learns highly interpretable representations from even only word annotations, and produces realistic sentences with desired attributes. Quantitative evaluation validates the accuracy of sentence and attribute generation.",
"Perception and expression of emotion are key factors to the success of dialogue systems or conversational agents. However, this problem has not been studied in large-scale conversation generation so far. In this paper, we propose Emotional Chatting Machine (ECM) that can generate appropriate responses not only in content (relevant and grammatical) but also in emotion (emotionally consistent). To the best of our knowledge, this is the first work that addresses the emotion factor in large-scale conversation generation. ECM addresses the factor using three new mechanisms that respectively (1) models the high-level abstraction of emotion expressions by embedding emotion categories, (2) captures the change of implicit internal emotion states, and (3) uses explicit emotion expressions with an external emotion vocabulary. Experiments show that the proposed model can generate responses appropriate not only in content but also in emotion."
]
}
|
1811.09789
|
2901974839
|
There has been much recent work on image captioning models that describe the factual aspects of an image. Recently, some models have incorporated non-factual aspects into the captions, such as sentiment or style. However, such models typically have difficulty in balancing the semantic aspects of the image and the non-factual dimensions of the caption; in addition, it can be observed that humans may focus on different aspects of an image depending on the chosen sentiment or style of the caption. To address this, we design an attention-based model to better add sentiment to image captions. The model embeds and learns sentiment with respect to image-caption data, and uses both high-level and word-level sentiment information during the learning process. The model outperforms the state-of-the-art work in image captioning with sentiment using standard evaluation metrics. An analysis of generated captions also shows that our model does this by a better selection of the sentiment-bearing adjectives and adjective-noun pairs.
|
Moreover, image captioning systems control sentiment or other non-factual characteristics of the generated captions @cite_28 @cite_6 . In addition to describing the visual content, these models learn to generate different forms of captions. For instance, Mathews al @cite_6 proposed system to generate sentimental captions. Here, the notion of sentiment is drawn from Natural Language Processing @cite_23 , with sentiment either or . The system of Mathews al @cite_6 is a full switching architecture incorporating both factual and sentimental caption paths. It needs two-stage training: training on factual image captions and training on sentimental image captions. Therefore, it does not support end-to-end training.
|
{
"cite_N": [
"@cite_28",
"@cite_23",
"@cite_6"
],
"mid": [
"2625940279",
"2097726431",
"2963062932"
],
"abstract": [
"We propose a novel framework named StyleNet to address the task of generating attractive captions for images and videos with different styles. To this end, we devise a novel model component, named factored LSTM, which automatically distills the style factors in the monolingual text corpus. Then at runtime, we can explicitly control the style in the caption generation process so as to produce attractive visual captions with the desired style. Our approach achieves this goal by leveraging two sets of data: 1) factual image video-caption paired data, and 2) stylized monolingual text data (e.g., romantic and humorous sentences). We show experimentally that StyleNet outperforms existing approaches for generating visual captions with different styles, measured in both automatic and human evaluation metrics on the newly collected FlickrStyle10K image caption dataset, which contains 10K Flickr images with corresponding humorous and romantic captions.",
"An important part of our information-gathering behavior has always been to find out what other people think. With the growing availability and popularity of opinion-rich resources such as online review sites and personal blogs, new opportunities and challenges arise as people now can, and do, actively use information technologies to seek out and understand the opinions of others. The sudden eruption of activity in the area of opinion mining and sentiment analysis, which deals with the computational treatment of opinion, sentiment, and subjectivity in text, has thus occurred at least in part as a direct response to the surge of interest in new systems that deal directly with opinions as a first-class object. This survey covers techniques and approaches that promise to directly enable opinion-oriented information-seeking systems. Our focus is on methods that seek to address the new challenges raised by sentiment-aware applications, as compared to those that are already present in more traditional fact-based analysis. We include material on summarization of evaluative text and on broader issues regarding privacy, manipulation, and economic impact that the development of opinion-oriented information-access services gives rise to. To facilitate future work, a discussion of available resources, benchmark datasets, and evaluation campaigns is also provided.",
"The recent progress on image recognition and language modeling is making automatic description of image content a reality. However, stylized, non-factual aspects of the written description are missing from the current systems. One such style is descriptions with emotions, which is commonplace in everyday communication, and influences decision-making and interpersonal relationships. We design a system to describe an image with emotions, and present a model that automatically generates captions with positive or negative sentiments. We propose a novel switching recurrent neural network with word-level regularization, which is able to produce emotional image captions using only 2000+ training sentences containing sentiments. We evaluate the captions with different automatic and crowd-sourcing metrics. Our model compares favourably in common quality metrics for image captioning. In 84.6 of cases the generated positive captions were judged as being at least as descriptive as the factual captions. Of these positive captions 88 were confirmed by the crowd-sourced workers as having the appropriate sentiment."
]
}
|
1811.09789
|
2901974839
|
There has been much recent work on image captioning models that describe the factual aspects of an image. Recently, some models have incorporated non-factual aspects into the captions, such as sentiment or style. However, such models typically have difficulty in balancing the semantic aspects of the image and the non-factual dimensions of the caption; in addition, it can be observed that humans may focus on different aspects of an image depending on the chosen sentiment or style of the caption. To address this, we design an attention-based model to better add sentiment to image captions. The model embeds and learns sentiment with respect to image-caption data, and uses both high-level and word-level sentiment information during the learning process. The model outperforms the state-of-the-art work in image captioning with sentiment using standard evaluation metrics. An analysis of generated captions also shows that our model does this by a better selection of the sentiment-bearing adjectives and adjective-noun pairs.
|
To address this issue, You al @cite_30 designed two new schemes, and , to better employ sentiment in generating image captions. For , an additional dimension was added to the input of a recurrent neural network (RNN) to express sentiment. A related idea was earlier proposed by Radford al @cite_13 who discovered a sentiment unit in a RNN-based system. In the model, the sentiment unit is injected at every time step of the generation process. The approach of You al @cite_30 injects the sentiment unit only at the initial time step of a designated sentiment cell trained in a similar learning fashion to the memory cell in a long short-term memory (LSTM) network. Similar to the work of You al @cite_30 , we have a single phase optimization for our image captioning model. In contrast, , and models apply visual features only in the initial time step of the LSTM.
|
{
"cite_N": [
"@cite_30",
"@cite_13"
],
"mid": [
"2786399470",
"2606347107"
],
"abstract": [
"Automatic image captioning has recently approached human-level performance due to the latest advances in computer vision and natural language understanding. However, most of the current models can only generate plain factual descriptions about the content of a given image. However, for human beings, image caption writing is quite flexible and diverse, where additional language dimensions, such as emotion, humor and language styles, are often incorporated to produce diverse, emotional, or appealing captions. In particular, we are interested in generating sentiment-conveying image descriptions, which has received little attention. The main challenge is how to effectively inject sentiments into the generated captions without altering the semantic matching between the visual content and the generated descriptions. In this work, we propose two different models, which employ different schemes for injecting sentiments into image captions. Compared with the few existing approaches, the proposed models are much simpler and yet more effective. The experimental results show that our model outperform the state-of-the-art models in generating sentimental (i.e., sentiment-bearing) image captions. In addition, we can also easily manipulate the model by assigning different sentiments to the testing image to generate captions with the corresponding sentiments.",
"We explore the properties of byte-level recurrent language models. When given sufficient amounts of capacity, training data, and compute time, the representations learned by these models include disentangled features corresponding to high-level concepts. Specifically, we find a single unit which performs sentiment analysis. These representations, learned in an unsupervised manner, achieve state of the art on the binary subset of the Stanford Sentiment Treebank. They are also very data efficient. When using only a handful of labeled examples, our approach matches the performance of strong baselines trained on full datasets. We also demonstrate the sentiment unit has a direct influence on the generative process of the model. Simply fixing its value to be positive or negative generates samples with the corresponding positive or negative sentiment."
]
}
|
1811.09789
|
2901974839
|
There has been much recent work on image captioning models that describe the factual aspects of an image. Recently, some models have incorporated non-factual aspects into the captions, such as sentiment or style. However, such models typically have difficulty in balancing the semantic aspects of the image and the non-factual dimensions of the caption; in addition, it can be observed that humans may focus on different aspects of an image depending on the chosen sentiment or style of the caption. To address this, we design an attention-based model to better add sentiment to image captions. The model embeds and learns sentiment with respect to image-caption data, and uses both high-level and word-level sentiment information during the learning process. The model outperforms the state-of-the-art work in image captioning with sentiment using standard evaluation metrics. An analysis of generated captions also shows that our model does this by a better selection of the sentiment-bearing adjectives and adjective-noun pairs.
|
However, recent state-of-the-art image captioning models usually apply visual features, which are spatial ones, at each time step of the LSTM @cite_19 @cite_10 @cite_16 @cite_24 . Nezami al @cite_11 also used spatial features at every step to generate more human-like captions, but for injecting facial expressions. In this work, therefore, we use an attention-based model to generate sentimental captions. In addition to this, we design a new injecting mechanism for sentiments to learn both high-level and word-level information. The high-level one involves injecting a sentiment vector to the LSTM. The fine-grained one involves using another sentiment vector to learn sentiment values appropriate at the word level.
|
{
"cite_N": [
"@cite_24",
"@cite_19",
"@cite_16",
"@cite_10",
"@cite_11"
],
"mid": [
"2950178297",
"2951590222",
"2963084599",
"2952469094",
"2952048081"
],
"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.",
"Top-down visual attention mechanisms have been used extensively in image captioning and visual question answering (VQA) to enable deeper image understanding through fine-grained analysis and even multiple steps of reasoning. In this work, we propose a combined bottom-up and top-down attention mechanism that enables attention to be calculated at the level of objects and other salient image regions. This is the natural basis for attention to be considered. Within our approach, the bottom-up mechanism (based on Faster R-CNN) proposes image regions, each with an associated feature vector, while the top-down mechanism determines feature weightings. Applying this approach to image captioning, our results on the MSCOCO test server establish a new state-of-the-art for the task, achieving CIDEr SPICE BLEU-4 scores of 117.9, 21.5 and 36.9, respectively. Demonstrating the broad applicability of the method, applying the same approach to VQA we obtain first place in the 2017 VQA Challenge.",
"Recently it has been shown that policy-gradient methods for reinforcement learning can be utilized to train deep end-to-end systems directly on non-differentiable metrics for the task at hand. In this paper we consider the problem of optimizing image captioning systems using reinforcement learning, and show that by carefully optimizing our systems using the test metrics of the MSCOCO task, significant gains in performance can be realized. Our systems are built using a new optimization approach that we call self-critical sequence training (SCST). SCST is a form of the popular REINFORCE algorithm that, rather than estimating a baseline to normalize the rewards and reduce variance, utilizes the output of its own test-time inference algorithm to normalize the rewards it experiences. Using this approach, estimating the reward signal (as actor-critic methods must do) and estimating normalization (as REINFORCE algorithms typically do) is avoided, while at the same time harmonizing the model with respect to its test-time inference procedure. Empirically we find that directly optimizing the CIDEr metric with SCST and greedy decoding at test-time is highly effective. Our results on the MSCOCO evaluation sever establish a new state-of-the-art on the task, improving the best result in terms of CIDEr from 104.9 to 114.7.",
"Attention-based neural encoder-decoder frameworks have been widely adopted for image captioning. Most methods force visual attention to be active for every generated word. However, the decoder likely requires little to no visual information from the image to predict non-visual words such as \"the\" and \"of\". Other words that may seem visual can often be predicted reliably just from the language model e.g., \"sign\" after \"behind a red stop\" or \"phone\" following \"talking on a cell\". In this paper, we propose a novel adaptive attention model with a visual sentinel. At each time step, our model decides whether to attend to the image (and if so, to which regions) or to the visual sentinel. The model decides whether to attend to the image and where, in order to extract meaningful information for sequential word generation. We test our method on the COCO image captioning 2015 challenge dataset and Flickr30K. Our approach sets the new state-of-the-art by a significant margin.",
"Image captioning is the process of generating a natural language description of an image. Most current image captioning models, however, do not take into account the emotional aspect of an image, which is very relevant to activities and interpersonal relationships represented therein. Towards developing a model that can produce human-like captions incorporating these, we use facial expression features extracted from images including human faces, with the aim of improving the descriptive ability of the model. In this work, we present two variants of our Face-Cap model, which embed facial expression features in different ways, to generate image captions. Using all standard evaluation metrics, our Face-Cap models outperform a state-of-the-art baseline model for generating image captions when applied to an image caption dataset extracted from the standard Flickr 30K dataset, consisting of around 11K images containing faces. An analysis of the captions finds that, perhaps surprisingly, the improvement in caption quality appears to come not from the addition of adjectives linked to emotional aspects of the images, but from more variety in the actions described in the captions."
]
}
|
1811.09944
|
2949299200
|
Audit logs serve as a critical component in the enterprise business systems that are used for auditing, storing, and tracking changes made to the data. However, audit logs are vulnerable to a series of attacks, which enable adversaries to tamper data and corresponding audit logs. In this paper, we present BlockAudit: a scalable and tamper-proof system that leverages the design properties of audit logs and security guarantees of blockchains to enable secure and trustworthy audit logs. Towards that, we construct the design schema of BlockAudit, and outline its operational procedures. We implement our design on Hyperledger and evaluate its performance in terms of latency, network size, and payload size. Our results show that conventional audit logs can seamlessly transition into BlockAudit to achieve higher security, integrity, and fault tolerance.
|
Audit Logs Schneier and Kelsey @cite_6 @cite_14 proposed a secure audit logging scheme capable of tamper detection even after compromise. However, their system requires the audit log entries to be generated prior to the attack. Moreover, their system does not provide an effective way to stop the attacker from deleting or appending audit records, which, in our case is easily spotted by BlockAudit . Snodgrass al @cite_11 proposed a trusted notary based tampering detection mechanism for RDBMS audit logs. In their scheme, a check field is stored within each tuple, and when a tuple is modified, RDBMS obtains a timestamp and computes a hash of the new data along with the timestamp. The hash values are then sent as a digital document to the notarization service which replies with a unique notary ID. The unique ID is stored in the tuple, and if attacker changes the data or timestamp, the ID received from the notary becomes inconsistent, which can be used for attack detection.
|
{
"cite_N": [
"@cite_14",
"@cite_6",
"@cite_11"
],
"mid": [
"2098721736",
"1987593503",
"2155018912"
],
"abstract": [
"In many real-world applications, sensitive information must be kept in log files on an untrusted machine. In the event that an attacker captures this machine, we would like to guarantee that he will gain little or no information from the log files and to limit his ability to corrupt the log files. We describe a computationally cheap method for making all log entries generated prior to the logging machine's compromise impossible for the attacker to read, and also impossible to undetectably modify or destroy.",
"In many real-world applications, sensitive information must be kept it log files on an untrusted machine. In the event that an attacker captures this machine, we would like to guarantee that he will gain little or no information from the log files and to limit his ability to corrupt the log files. We describe a computationally cheap method for making all log entries generated prior to the logging machine's compromise impossible for the attacker to read, and also impossible to modify or destroy undetectably.",
"Audit logs are considered good practice for business systems, and are required by federal regulations for secure systems, drug approval data, medical information disclosure, financial records, and electronic voting. Given the central role of audit logs, it is critical that they are correct and inalterable. It is not sufficient to say, \"our data is correct, because we store all interactions in a separate audit log.\" The integrity of the audit log itself must also be guaranteed. This paper proposes mechanisms within a database management system (DBMS), based on cryptographically strong one-way hash functions, that prevent an intruder, including an auditor or an employee or even an unknown bug within the DBMS itself, from silently corrupting the audit log. We propose that the DBMS store additional information in the database to enable a separate audit log validator to examine the database along with this extra information and state conclusively whether the audit log has been compromised. We show with an implementation on a high-performance storage engine that the overhead for auditing is low and that the validator can efficiently and correctly determine if the audit log has been compromised."
]
}
|
1811.09944
|
2949299200
|
Audit logs serve as a critical component in the enterprise business systems that are used for auditing, storing, and tracking changes made to the data. However, audit logs are vulnerable to a series of attacks, which enable adversaries to tamper data and corresponding audit logs. In this paper, we present BlockAudit: a scalable and tamper-proof system that leverages the design properties of audit logs and security guarantees of blockchains to enable secure and trustworthy audit logs. Towards that, we construct the design schema of BlockAudit, and outline its operational procedures. We implement our design on Hyperledger and evaluate its performance in terms of latency, network size, and payload size. Our results show that conventional audit logs can seamlessly transition into BlockAudit to achieve higher security, integrity, and fault tolerance.
|
Blockchain and Audit Logs Sutton and Samvi @cite_17 proposed a blockchain-based approach that stores the integrity proof digest to the Bitcoin blockchain. Castaldo al @cite_20 proposed a logging system to facilitate the exchange of electronic health data across multiple countries in Europe. Cucrull al @cite_2 proposed a system that uses blockchains to enhances the security of the immutable logs. Log integrity proofs are published in the blockchain, and provide non-repudiation security properties resilient to log truncation and log regeneration. In contrast, BlockAudit generates audit logs by extending the existing ORM (nHibernate), which is localized to ORM and other layers of business applicaiton are not effected. This makes it straightforward for existing application to user BlockAudit .
|
{
"cite_N": [
"@cite_2",
"@cite_20",
"@cite_17"
],
"mid": [
"2520906649",
"2883678950",
"2763836263"
],
"abstract": [
"Several applications require robust and tamper-proof logging systems, e.g. electronic voting or bank information systems. At Scytl we use a technology, called immutable logs, that we deploy in our electronic voting solutions. This technology ensures the integrity, authenticity and non-repudiation of the generated logs, thus in case of any event the auditors can use them to investigate the issue. As a security recommendation it is advisable to store and or replicate the information logged in a location where the logger has no writing or modification permissions. Otherwise, if the logger gets compromised, the data previously generated could be truncated or altered using the same private keys. This approach is costly and does not protect against collusion between the logger and the entities that hold the replicated data. In order to tackle these issues, in this article we present a proposal and implementation to immutabilize integrity proofs of the secure logs within the Bitcoin’s blockchain. Due to the properties of the proposal, the integrity of the immutabilized logs is guaranteed without performing log data replication and even in case the logger gets latterly compromised.",
"On an EU level, the topic of electronic health data is a high priority. Many projects have been developed to realise a standard health data format to share information on a regional, national or EU level. All the projects favour and contribute to the development and improvement of the prerequisites for intra- and cross-border patient mobility. This work presents a new approach for the implementation of disruptive logging: an audit mechanism for cross-border exchange of eHealth data on OpenNCP, providing traceability and liability support within the OpenNCP infrastructure. Relevant parties could be legally obliged to keep a log of all privacy-critical operations performed by OpenNCP users.",
"Privacy audit logs are used to capture the actions of participants in a data sharing environment in order for auditors to check compliance with privacy policies. However, collusion may occur between the auditors and participants to obfuscate actions that should be recorded in the audit logs. In this paper, we propose a Linked Data based method of utilizing blockchain technology to create tamper-proof audit logs that provide proof of log manipulation and non-repudiation. We also provide experimental validation of the scalability of our solution using an existing Linked Data privacy audit log model."
]
}
|
1811.09855
|
2901716381
|
This paper investigates how to perform robust visual tracking in adverse and challenging conditions using complementary visual and thermal infrared data (RGB-T tracking). We propose a novel deep network architecture "quality-aware Feature Aggregation Network (FANet)" to achieve quality-aware aggregations of both hierarchical features and multimodal information for robust online RGB-T tracking. Unlike existing works that directly concatenate hierarchical deep features, our FANet learns the layer weights to adaptively aggregate them to handle the challenge of significant appearance changes caused by deformation, abrupt motion, background clutter and occlusion within each modality. Moreover, we employ the operations of max pooling, interpolation upsampling and convolution to transform these hierarchical and multi-resolution features into a uniform space at the same resolution for more effective feature aggregation. In different modalities, we elaborately design a multimodal aggregation sub-network to integrate all modalities collaboratively based on the predicted reliability degrees. Extensive experiments on large-scale benchmark datasets demonstrate that our FANet significantly outperforms other state-of-the-art RGB-T tracking methods.
|
MDNet @cite_31 achieved the state-of-the-art performance on multiple datasets by dealing with label conflict issue across videos through multi-domain learning. Han @cite_11 proposed to select a subset of branches in the CNN randomly for online learning whenever target appearance models need to be updated for better regularization, where each branch may have a different number of layers to maintain variable abstraction levels of target appearances. The meta learning was introduced in MDNet to adjust the initial deep networks @cite_29 , which could quickly adapt to robustly model a particular target in future frames. Jung @cite_12 proposed a real-time MDNet, where an improved RoIAlign technique was employed to extract more accurate representations for targets. All these methods were developed for single-modality tracking and we study them in the task of multi-modality tracking in this work.
|
{
"cite_N": [
"@cite_31",
"@cite_29",
"@cite_12",
"@cite_11"
],
"mid": [
"1857884451",
"2783173047",
"",
"2737572441"
],
"abstract": [
"We propose a novel visual tracking algorithm based on the representations from a discriminatively trained Convolutional Neural Network (CNN). Our algorithm pretrains a CNN using a large set of videos with tracking groundtruths to obtain a generic target representation. Our network is composed of shared layers and multiple branches of domain-specific layers, where domains correspond to individual training sequences and each branch is responsible for binary classification to identify target in each domain. We train each domain in the network iteratively to obtain generic target representations in the shared layers. When tracking a target in a new sequence, we construct a new network by combining the shared layers in the pretrained CNN with a new binary classification layer, which is updated online. Online tracking is performed by evaluating the candidate windows randomly sampled around the previous target state. The proposed algorithm illustrates outstanding performance in existing tracking benchmarks.",
"This paper improves state-of-the-art on-line trackers that use deep learning. Such trackers train a deep network to pick a specified object out from the background in an initial frame (initialization) and then keep training the model as tracking proceeds (updates). Our core contribution is a meta-learning-based method to adjust deep networks for tracking using off-line training. First, we learn initial parameters and per-parameter coefficients for fast online adaptation. Second, we use training signal from future frames for robustness to target appearance variations and environment changes. The resulting networks train significantly faster during the initialization, while improving robustness and accuracy. We demonstrate this approach on top of the current highest accuracy tracking approach, tracking-by-detection based MDNet and close competitor, the correlation-based CREST. Experimental results on both standard benchmarks, OTB and VOT2016, show improvements in speed, accuracy, and robustness on both trackers.",
"",
"We propose an extremely simple but effective regularization technique of convolutional neural networks (CNNs), referred to as BranchOut, for online ensemble tracking. Our algorithm employs a CNN for target representation, which has a common convolutional layers but has multiple branches of fully connected layers. For better regularization, a subset of branches in the CNN are selected randomly for online learning whenever target appearance models need to be updated. Each branch may have a different number of layers to maintain variable abstraction levels of target appearances. BranchOut with multi-level target representation allows us to learn robust target appearance models with diversity and handle various challenges in visual tracking problem effectively. The proposed algorithm is evaluated in standard tracking benchmarks and shows the state-of-the-art performance even without additional pretraining on external tracking sequences."
]
}
|
1811.09791
|
2901382504
|
In this paper, we address the problem of unsupervised video summarization that automatically extracts key-shots from an input video. Specifically, we tackle two critical issues based on our empirical observations: (i) Ineffective feature learning due to flat distributions of output importance scores for each frame, and (ii) training difficulty when dealing with long-length video inputs. To alleviate the first problem, we propose a simple yet effective regularization loss term called variance loss. The proposed variance loss allows a network to predict output scores for each frame with high discrepancy which enables effective feature learning and significantly improves model performance. For the second problem, we design a novel two-stream network named Chunk and Stride Network (CSNet) that utilizes local (chunk) and global (stride) temporal view on the video features. Our CSNet gives better summarization results for long-length videos compared to the existing methods. In addition, we introduce an attention mechanism to handle the dynamic information in videos. We demonstrate the effectiveness of the proposed methods by conducting extensive ablation studies and show that our final model achieves new state-of-the-art results on two benchmark datasets.
|
Given an input video, video summarization aims to produce a shortened version that highlights the representative video frames. Various prior work has proposed solutions to this problem, including video time-lapse @cite_11 @cite_22 @cite_2 , synopsis @cite_32 , montage @cite_3 @cite_6 and storyboards @cite_23 @cite_1 @cite_27 @cite_31 @cite_24 @cite_28 @cite_23 . Our work is most closely related to storyboards, selecting some important pieces of information to summarize key events present in the entire video.
|
{
"cite_N": [
"@cite_22",
"@cite_28",
"@cite_1",
"@cite_32",
"@cite_3",
"@cite_6",
"@cite_24",
"@cite_27",
"@cite_23",
"@cite_2",
"@cite_31",
"@cite_11"
],
"mid": [
"",
"2952694903",
"2529272619",
"2115060048",
"",
"333230188",
"2116946038",
"1904325426",
"2115857089",
"1948812921",
"2106229755",
"2003553461"
],
"abstract": [
"",
"With the growing popularity of short-form video sharing platforms such as Instagram and Vine , there has been an increasing need for techniques that automatically extract highlights from video. Whereas prior works have approached this problem with heuristic rules or supervised learning, we present an unsupervised learning approach that takes advantage of the abundance of user-edited videos on social media websites such as YouTube. Based on the idea that the most significant sub-events within a video class are commonly present among edited videos while less interesting ones appear less frequently, we identify the significant sub-events via a robust recurrent auto-encoder trained on a collection of user-edited videos queried for each particular class of interest. The auto-encoder is trained using a proposed shrinking exponential loss function that makes it robust to noise in the web-crawled training data, and is configured with bidirectional long short term memory (LSTM) LSTM:97 cells to better model the temporal structure of highlight segments. Different from supervised techniques, our method can infer highlights using only a set of downloaded edited videos, without also needing their pre-edited counterparts which are rarely available online. Extensive experiments indicate the promise of our proposed solution in this challenging unsupervised settin",
"This paper proposes a novel approach and a new benchmark for video summarization. Thereby we focus on user videos, which are raw videos containing a set of interesting events. Our method starts by segmenting the video by using a novel “superframe” segmentation, tailored to raw videos. Then, we estimate visual interestingness per superframe using a set of low-, mid- and high-level features. Based on this scoring, we select an optimal subset of superframes to create an informative and interesting summary. The introduced benchmark comes with multiple human created summaries, which were acquired in a controlled psychological experiment. This data paves the way to evaluate summarization methods objectively and to get new insights in video summarization. When evaluating our method, we find that it generates high-quality results, comparable to manual, human-created summaries.",
"The amount of captured video is growing with the increased numbers of video cameras, especially the increase of millions of surveillance cameras that operate 24 hours a day. Since video browsing and retrieval is time consuming, most captured video is never watched or examined. Video synopsis is an effective tool for browsing and indexing of such a video. It provides a short video representation, while preserving the essential activities of the original video. The activity in the video is condensed into a shorter period by simultaneously showing multiple activities, even when they originally occurred at different times. The synopsis video is also an index into the original video by pointing to the original time of each activity. Video synopsis can be applied to create a synopsis of an endless video streams, as generated by Webcams and by surveillance cameras. It can address queries like \"show in one minute the synopsis of this camera broadcast during the past day''. This process includes two major phases: (i) an online conversion of the endless video stream into a database of objects and activities (rather than frames). (ii) A response phase, generating the video synopsis as a response to the user's query.",
"",
"We present a novel method to generate salient montages from unconstrained videos, by finding “montageable moments” and identifying the salient people and actions to depict in each montage. Our method addresses the need for generating concise visualizations from the increasingly large number of videos being captured from portable devices. Our main contributions are (1) the process of finding salient people and moments to form a montage, and (2) the application of this method to videos taken “in the wild” where the camera moves freely. As such, we demonstrate results on head-mounted cameras, where the camera moves constantly, as well as on videos downloaded from YouTube. Our approach can operate on videos of any length; some will contain many montageable moments, while others may have none. We demonstrate that a novel “montageability” score can be used to retrieve results with relatively high precision which allows us to present high quality montages to users.",
"We propose a novel method for removing irrelevant frames from a video given user-provided frame-level labeling for a very small number of frames. We first hypothesize a number of windows which possibly contain the object of interest, and then determine which window(s) truly contain the object of interest. Our method enjoys several favorable properties. First, compared to approaches where a single descriptor is used to describe a whole frame, each window's feature descriptor has the chance of genuinely describing the object of interest; hence it is less affected by background clutter. Second, by considering the temporal continuity of a video instead of treating frames as independent, we can hypothesize the location of the windows more accurately. Third, by infusing prior knowledge into the patch-level model, we can precisely follow the trajectory of the object of interest. This allows us to largely reduce the number of windows and hence reduce the chance of overfitting the data during learning. We demonstrate the effectiveness of the method by comparing it to several other semi-supervised learning approaches on challenging video clips.",
"We present a novel method for summarizing raw, casually captured videos. The objective is to create a short summary that still conveys the story. It should thus be both, interesting and representative for the input video. Previous methods often used simplified assumptions and only optimized for one of these goals. Alternatively, they used handdefined objectives that were optimized sequentially by making consecutive hard decisions. This limits their use to a particular setting. Instead, we introduce a new method that (i) uses a supervised approach in order to learn the importance of global characteristics of a summary and (ii) jointly optimizes for multiple objectives and thus creates summaries that posses multiple properties of a good summary. Experiments on two challenging and very diverse datasets demonstrate the effectiveness of our method, where we outperform or match current state-of-the-art.",
"Video summarization is a challenging problem with great application potential. Whereas prior approaches, largely unsupervised in nature, focus on sampling useful frames and assembling them as summaries, we consider video summarization as a supervised subset selection problem. Our idea is to teach the system to learn from human-created summaries how to select informative and diverse subsets, so as to best meet evaluation metrics derived from human-perceived quality. To this end, we propose the sequential determinantal point process (seqDPP), a probabilistic model for diverse sequential subset selection. Our novel seqDPP heeds the inherent sequential structures in video data, thus overcoming the deficiency of the standard DPP, which treats video frames as randomly permutable items. Meanwhile, seqDPP retains the power of modeling diverse subsets, essential for summarization. Our extensive results of summarizing videos from 3 datasets demonstrate the superior performance of our method, compared to not only existing unsupervised methods but also naive applications of the standard DPP model.",
"While egocentric cameras like GoPro are gaining popularity, the videos they capture are long, boring, and difficult to watch from start to end. Fast forwarding (i.e. frame sampling) is a natural choice for faster video browsing. However, this accentuates the shake caused by natural head motion, making the fast forwarded video useless.",
"We present a video summarization approach for egocentric or “wearable” camera data. Given hours of video, the proposed method produces a compact storyboard summary of the camera wearer's day. In contrast to traditional keyframe selection techniques, the resulting summary focuses on the most important objects and people with which the camera wearer interacts. To accomplish this, we develop region cues indicative of high-level saliency in egocentric video — such as the nearness to hands, gaze, and frequency of occurrence — and learn a regressor to predict the relative importance of any new region based on these cues. Using these predictions and a simple form of temporal event detection, our method selects frames for the storyboard that reflect the key object-driven happenings. Critically, the approach is neither camera-wearer-specific nor object-specific; that means the learned importance metric need not be trained for a given user or context, and it can predict the importance of objects and people that have never been seen previously. Our results with 17 hours of egocentric data show the method's promise relative to existing techniques for saliency and summarization.",
"Long videos can be played much faster than real-time by recording only one frame per second or by dropping all but one frame each second, i.e., by creating a timelapse. Unstable hand-held moving videos can be stabilized with a number of recently described methods. Unfortunately, creating a stabilized timelapse, or hyperlapse, cannot be achieved through a simple combination of these two methods. Two hyperlapse methods have been previously demonstrated: one with high computational complexity and one requiring special sensors. We present an algorithm for creating hyperlapse videos that can handle significant high-frequency camera motion and runs in real-time on HD video. Our approach does not require sensor data, thus can be run on videos captured on any camera. We optimally select frames from the input video that best match a desired target speed-up while also resulting in the smoothest possible camera motion. We evaluate our approach using several input videos from a range of cameras and compare these results to existing methods."
]
}
|
1811.09791
|
2901382504
|
In this paper, we address the problem of unsupervised video summarization that automatically extracts key-shots from an input video. Specifically, we tackle two critical issues based on our empirical observations: (i) Ineffective feature learning due to flat distributions of output importance scores for each frame, and (ii) training difficulty when dealing with long-length video inputs. To alleviate the first problem, we propose a simple yet effective regularization loss term called variance loss. The proposed variance loss allows a network to predict output scores for each frame with high discrepancy which enables effective feature learning and significantly improves model performance. For the second problem, we design a novel two-stream network named Chunk and Stride Network (CSNet) that utilizes local (chunk) and global (stride) temporal view on the video features. Our CSNet gives better summarization results for long-length videos compared to the existing methods. In addition, we introduce an attention mechanism to handle the dynamic information in videos. We demonstrate the effectiveness of the proposed methods by conducting extensive ablation studies and show that our final model achieves new state-of-the-art results on two benchmark datasets.
|
Early work on video summarization problems heavily relied on hand-crafted features and unsupervised learning. Such work defined various heuristics to represent the importance of the frames @cite_20 @cite_25 @cite_10 @cite_29 @cite_15 and to use the scores to select representative frames to build the summary video. Recent work has explored supervised learning approach for this problem, using training data consisting of videos and their ground-truth summaries generated by humans. These supervised learning methods outperform early work on unsupervised approach, since they can better learn the high-level semantic knowledge that is used by humans to generate summaries.
|
{
"cite_N": [
"@cite_29",
"@cite_15",
"@cite_10",
"@cite_25",
"@cite_20"
],
"mid": [
"1984899418",
"2103908291",
"2120645068",
"2109152179",
"1924343884"
],
"abstract": [
"In this paper, we investigate an approach for reconstructing storyline graphs from large-scale collections of Internet images, and optionally other side information such as friendship graphs. The storyline graphs can be an effective summary that visualizes various branching narrative structure of events or activities recurring across the input photo sets of a topic class. In order to explore further the usefulness of the storyline graphs, we leverage them to perform the image sequential prediction tasks, from which photo recommendation applications can benefit. We formulate the storyline reconstruction problem as an inference of sparse time-varying directed graphs, and develop an optimization algorithm that successfully addresses a number of key challenges of Web-scale problems, including global optimality, linear complexity, and easy parallelization. With experiments on more than 3.3 millions of images of 24 classes and user studies via Amazon Mechanical Turk, we show that the proposed algorithm improves other candidate methods for both storyline reconstruction and image prediction tasks.",
"Given the enormous growth in user-generated videos, it is becoming increasingly important to be able to navigate them efficiently. As these videos are generally of poor quality, summarization methods designed for well-produced videos do not generalize to them. To address this challenge, we propose to use web-images as a prior to facilitate summarization of user-generated videos. Our main intuition is that people tend to take pictures of objects to capture them in a maximally informative way. Such images could therefore be used as prior information to summarize videos containing a similar set of objects. In this work, we apply our novel insight to develop a summarization algorithm that uses the web-image based prior information in an unsupervised manner. Moreover, to automatically evaluate summarization algorithms on a large scale, we propose a framework that relies on multiple summaries obtained through crowdsourcing. We demonstrate the effectiveness of our evaluation framework by comparing its performance to that of multiple human evaluators. Finally, we present results for our framework tested on hundreds of user-generated videos.",
"We present a video summarization approach that discovers the story of an egocentric video. Given a long input video, our method selects a short chain of video sub shots depicting the essential events. Inspired by work in text analysis that links news articles over time, we define a random-walk based metric of influence between sub shots that reflects how visual objects contribute to the progression of events. Using this influence metric, we define an objective for the optimal k-subs hot summary. Whereas traditional methods optimize a summary's diversity or representative ness, ours explicitly accounts for how one sub-event \"leads to\" another-which, critically, captures event connectivity beyond simple object co-occurrence. As a result, our summaries provide a better sense of story. We apply our approach to over 12 hours of daily activity video taken from 23 unique camera wearers, and systematically evaluate its quality compared to multiple baselines with 34 human subjects.",
"We propose a unified approach for summarization based on the analysis of video structures and video highlights. Our approach emphasizes both the content balance and perceptual quality of a summary. Normalized cut algorithm is employed to globally and optimally partition a video into clusters. A motion attention model based on human perception is employed to compute the perceptual quality of shots and clusters. The clusters, together with the computed attention values, form a temporal graph similar to Markov chain that inherently describes the evolution and perceptual importance of video clusters. In our application, the flow of a temporal graph is utilized to group similar clusters into scenes, while the attention values are used as guidelines to select appropriate subshots in scenes for summarization.",
"Video summarization is a challenging problem in part because knowing which part of a video is important requires prior knowledge about its main topic. We present TVSum, an unsupervised video summarization framework that uses title-based image search results to find visually important shots. We observe that a video title is often carefully chosen to be maximally descriptive of its main topic, and hence images related to the title can serve as a proxy for important visual concepts of the main topic. However, because titles are free-formed, unconstrained, and often written ambiguously, images searched using the title can contain noise (images irrelevant to video content) and variance (images of different topics). To deal with this challenge, we developed a novel co-archetypal analysis technique that learns canonical visual concepts shared between video and images, but not in either alone, by finding a joint-factorial representation of two data sets. We introduce a new benchmark dataset, TVSum50, that contains 50 videos and their shot-level importance scores annotated via crowdsourcing. Experimental results on two datasets, SumMe and TVSum50, suggest our approach produces superior quality summaries compared to several recently proposed approaches."
]
}
|
1811.09791
|
2901382504
|
In this paper, we address the problem of unsupervised video summarization that automatically extracts key-shots from an input video. Specifically, we tackle two critical issues based on our empirical observations: (i) Ineffective feature learning due to flat distributions of output importance scores for each frame, and (ii) training difficulty when dealing with long-length video inputs. To alleviate the first problem, we propose a simple yet effective regularization loss term called variance loss. The proposed variance loss allows a network to predict output scores for each frame with high discrepancy which enables effective feature learning and significantly improves model performance. For the second problem, we design a novel two-stream network named Chunk and Stride Network (CSNet) that utilizes local (chunk) and global (stride) temporal view on the video features. Our CSNet gives better summarization results for long-length videos compared to the existing methods. In addition, we introduce an attention mechanism to handle the dynamic information in videos. We demonstrate the effectiveness of the proposed methods by conducting extensive ablation studies and show that our final model achieves new state-of-the-art results on two benchmark datasets.
|
Recently, deep learning based methods @cite_33 @cite_0 @cite_16 have gained attention for video summarization tasks. The most recent studies adopt recurrent models such as LSTMs, based on the intuition that using LSTM enables the capture of long-range temporal dependencies among video frames which are critical for effective summary generation.
|
{
"cite_N": [
"@cite_0",
"@cite_16",
"@cite_33"
],
"mid": [
"2737677090",
"",
"2963919999"
],
"abstract": [
"This paper addresses the problem of unsupervised video summarization, formulated as selecting a sparse subset of video frames that optimally represent the input video. Our key idea is to learn a deep summarizer network to minimize distance between training videos and a distribution of their summarizations, in an unsupervised way. Such a summarizer can then be applied on a new video for estimating its optimal summarization. For learning, we specify a novel generative adversarial framework, consisting of the summarizer and discriminator. The summarizer is the autoencoder long short-term memory network (LSTM) aimed at, first, selecting video frames, and then decoding the obtained summarization for reconstructing the input video. The discriminator is another LSTM aimed at distinguishing between the original video and its reconstruction from the summarizer. The summarizer LSTM is cast as an adversary of the discriminator, i.e., trained so as to maximally confuse the discriminator. This learning is also regularized for sparsity. Evaluation on four benchmark datasets, consisting of videos showing diverse events in first-and third-person views, demonstrates our competitive performance in comparison to fully supervised state-of-the-art approaches.",
"",
"We propose a novel supervised learning technique for summarizing videos by automatically selecting keyframes or key subshots. Casting the task as a structured prediction problem, our main idea is to use Long Short-Term Memory (LSTM) to model the variable-range temporal dependency among video frames, so as to derive both representative and compact video summaries. The proposed model successfully accounts for the sequential structure crucial to generating meaningful video summaries, leading to state-of-the-art results on two benchmark datasets. In addition to advances in modeling techniques, we introduce a strategy to address the need for a large amount of annotated data for training complex learning approaches to summarization. There, our main idea is to exploit auxiliary annotated video summarization datasets, in spite of their heterogeneity in visual styles and contents. Specifically, we show that domain adaptation techniques can improve learning by reducing the discrepancies in the original datasets’ statistical properties."
]
}
|
1811.09791
|
2901382504
|
In this paper, we address the problem of unsupervised video summarization that automatically extracts key-shots from an input video. Specifically, we tackle two critical issues based on our empirical observations: (i) Ineffective feature learning due to flat distributions of output importance scores for each frame, and (ii) training difficulty when dealing with long-length video inputs. To alleviate the first problem, we propose a simple yet effective regularization loss term called variance loss. The proposed variance loss allows a network to predict output scores for each frame with high discrepancy which enables effective feature learning and significantly improves model performance. For the second problem, we design a novel two-stream network named Chunk and Stride Network (CSNet) that utilizes local (chunk) and global (stride) temporal view on the video features. Our CSNet gives better summarization results for long-length videos compared to the existing methods. In addition, we introduce an attention mechanism to handle the dynamic information in videos. We demonstrate the effectiveness of the proposed methods by conducting extensive ablation studies and show that our final model achieves new state-of-the-art results on two benchmark datasets.
|
Zhang al @cite_33 introduced two LSTMs to model the variable range dependency in video summarization. One LSTM was used for video frame sequences in the forward direction, while the other LSTM was used for the backward direction. In addition, a determinantal point process model @cite_23 @cite_7 was adopted for further improvement of diversity in the subset selection. Mahasseni al. @cite_0 proposed an unsupervised method that was based on a generative adversarial framework. The model consists of the summarizer and discriminator. The summarizer was a variational autoencoder LSTM, which first summarized video and then reconstructed the output. The discriminator was another LSTM that learned to distinguish between its reconstruction and the input video.
|
{
"cite_N": [
"@cite_0",
"@cite_7",
"@cite_33",
"@cite_23"
],
"mid": [
"2737677090",
"2296744054",
"2963919999",
"2115857089"
],
"abstract": [
"This paper addresses the problem of unsupervised video summarization, formulated as selecting a sparse subset of video frames that optimally represent the input video. Our key idea is to learn a deep summarizer network to minimize distance between training videos and a distribution of their summarizations, in an unsupervised way. Such a summarizer can then be applied on a new video for estimating its optimal summarization. For learning, we specify a novel generative adversarial framework, consisting of the summarizer and discriminator. The summarizer is the autoencoder long short-term memory network (LSTM) aimed at, first, selecting video frames, and then decoding the obtained summarization for reconstructing the input video. The discriminator is another LSTM aimed at distinguishing between the original video and its reconstruction from the summarizer. The summarizer LSTM is cast as an adversary of the discriminator, i.e., trained so as to maximally confuse the discriminator. This learning is also regularized for sparsity. Evaluation on four benchmark datasets, consisting of videos showing diverse events in first-and third-person views, demonstrates our competitive performance in comparison to fully supervised state-of-the-art approaches.",
"Video summarization has unprecedented importance to help us digest, browse, and search today's ever-growing video collections. We propose a novel subset selection technique that leverages supervision in the form of human-created summaries to perform automatic keyframe-based video summarization. The main idea is to nonparametrically transfer summary structures from annotated videos to unseen test videos. We show how to extend our method to exploit semantic side information about the video's category genre to guide the transfer process by those training videos semantically consistent with the test input. We also show how to generalize our method to subshot-based summarization, which not only reduces computational costs but also provides more flexible ways of defining visual similarity across subshots spanning several frames. We conduct extensive evaluation on several benchmarks and demonstrate promising results, outperforming existing methods in several settings.",
"We propose a novel supervised learning technique for summarizing videos by automatically selecting keyframes or key subshots. Casting the task as a structured prediction problem, our main idea is to use Long Short-Term Memory (LSTM) to model the variable-range temporal dependency among video frames, so as to derive both representative and compact video summaries. The proposed model successfully accounts for the sequential structure crucial to generating meaningful video summaries, leading to state-of-the-art results on two benchmark datasets. In addition to advances in modeling techniques, we introduce a strategy to address the need for a large amount of annotated data for training complex learning approaches to summarization. There, our main idea is to exploit auxiliary annotated video summarization datasets, in spite of their heterogeneity in visual styles and contents. Specifically, we show that domain adaptation techniques can improve learning by reducing the discrepancies in the original datasets’ statistical properties.",
"Video summarization is a challenging problem with great application potential. Whereas prior approaches, largely unsupervised in nature, focus on sampling useful frames and assembling them as summaries, we consider video summarization as a supervised subset selection problem. Our idea is to teach the system to learn from human-created summaries how to select informative and diverse subsets, so as to best meet evaluation metrics derived from human-perceived quality. To this end, we propose the sequential determinantal point process (seqDPP), a probabilistic model for diverse sequential subset selection. Our novel seqDPP heeds the inherent sequential structures in video data, thus overcoming the deficiency of the standard DPP, which treats video frames as randomly permutable items. Meanwhile, seqDPP retains the power of modeling diverse subsets, essential for summarization. Our extensive results of summarizing videos from 3 datasets demonstrate the superior performance of our method, compared to not only existing unsupervised methods but also naive applications of the standard DPP model."
]
}
|
1811.09950
|
2901869505
|
Computer-vision hospital systems can greatly assist healthcare workers and improve medical facility treatment, but often face patient resistance due to the perceived intrusiveness and violation of privacy associated with visual surveillance. We downsample video frames to extremely low resolutions to degrade private information from surveillance videos. We measure the amount of activity-recognition information retained in low resolution depth images, and also apply a privately-trained DCSCN super-resolution model to enhance the utility of our images. We implement our techniques with two actual healthcare-surveillance scenarios, hand-hygiene compliance and ICU activity-logging, and show that our privacy-preserving techniques preserve enough information for realistic healthcare tasks.
|
Several works also explore low-resolution facial recognition which could be applicable to a privacy-sensitive context. One such work @cite_34 attempts to learn a common feature space between low and high resolution images using center regularization and GAN-based techniques. Another @cite_2 uses a two branch network to learn a cross representation between low and high-resolution faces.
|
{
"cite_N": [
"@cite_34",
"@cite_2"
],
"mid": [
"2805283564",
"2712273292"
],
"abstract": [
"Although face recognition systems have achieved impressive performance in recent years, the low-resolution face recognition (LRFR) task remains challenging, especially when the LR faces are captured under non-ideal conditions, as is common in surveillance-based applications. Faces captured in such conditions are often contaminated by blur, nonuniform lighting, and nonfrontal face pose. In this paper, we analyze face recognition techniques using data captured under low-quality conditions in the wild. We provide a comprehensive analysis of experimental results for two of the most important applications in real surveillance applications, and demonstrate practical approaches to handle both cases that show promising performance. The following three contributions are made: (i) we conduct experiments to evaluate super-resolution methods for low-resolution face recognition; (ii) we study face re-identification on various public face datasets including real surveillance and low-resolution subsets of large-scale datasets, present a baseline result for several deep learning based approaches, and improve them by introducing a GAN pre-training approach and fully convolutional architecture; and (iii) we explore low-resolution face identification by employing a state-of-the-art supervised discriminative learning approach. Evaluations are conducted on challenging portions of the SCFace and UCCSface datasets.",
"We propose a novel couple mappings method for low resolution face recognition using deep convolutional neural networks (DCNNs). The proposed architecture consists of two branches of DCNNs to map the high and low resolution face images into a common space with nonlinear transformations. The branch corresponding to transformation of high resolution images consists of 14 layers and the other branch which maps the low resolution face images to the common space includes a 5-layer super-resolution network connected to a 14-layer network. The distance between the features of corresponding high and low resolution images are backpropagated to train the networks. Our proposed method is evaluated on FERET data set and compared with state-of-the-art competing methods. Our extensive experimental results show that the proposed method significantly improves the recognition performance especially for very low resolution probe face images (11.4 improvement in recognition accuracy). Furthermore, it can reconstruct a high resolution image from its corresponding low resolution probe image which is comparable with state-of-the-art super-resolution methods in terms of visual quality."
]
}
|
1811.09725
|
2901616798
|
Deep learning is currently playing a crucial role toward higher levels of artificial intelligence. This paradigm allows neural networks to learn complex and abstract representations, that are progressively obtained by combining simpler ones. Nevertheless, the internal "black-box" representations automatically discovered by current neural architectures often suffer from a lack of interpretability, making of primary interest the study of explainable machine learning techniques. This paper summarizes our recent efforts to develop a more interpretable neural model for directly processing speech from the raw waveform. In particular, we propose SincNet, a novel Convolutional Neural Network (CNN) that encourages the first layer to discover more meaningful filters by exploiting parametrized sinc functions. In contrast to standard CNNs, which learn all the elements of each filter, only low and high cutoff frequencies of band-pass filters are directly learned from data. This inductive bias offers a very compact way to derive a customized filter-bank front-end, that only depends on some parameters with a clear physical meaning. Our experiments, conducted on both speaker and speech recognition, show that the proposed architecture converges faster, performs better, and is more interpretable than standard CNNs.
|
Several works have recently explored the use of low-level speech representations to process audio and speech with CNNs. Most prior attempts exploit magnitude spectrogram features @cite_29 @cite_9 @cite_10 @cite_59 @cite_47 @cite_28 . Although spectrograms retain more information than standard hand-crafted features, their design still requires careful tuning of some crucial hyper-parameters, such as the duration, overlap, and typology of the frame window, as well as the number of frequency bins. For this reason, a more recent trend is to directly learn from raw waveforms, thus completely avoiding any feature extraction step. This approach has shown promise in speech @cite_53 @cite_11 @cite_7 @cite_64 @cite_20 , including emotion tasks @cite_30 , speaker recognition @cite_60 , spoofing detection @cite_48 , and speech synthesis @cite_42 @cite_2 .
|
{
"cite_N": [
"@cite_30",
"@cite_64",
"@cite_7",
"@cite_60",
"@cite_28",
"@cite_48",
"@cite_9",
"@cite_29",
"@cite_53",
"@cite_42",
"@cite_59",
"@cite_2",
"@cite_47",
"@cite_10",
"@cite_20",
"@cite_11"
],
"mid": [
"2399733683",
"",
"1542280630",
"2770454110",
"2658929981",
"2592641653",
"2746742816",
"2802973008",
"1666984270",
"2519091744",
"1969851134",
"2584032004",
"2587891104",
"2964296349",
"2408093180",
"2398826216"
],
"abstract": [
"The automatic recognition of spontaneous emotions from speech is a challenging task. On the one hand, acoustic features need to be robust enough to capture the emotional content for various styles of speaking, and while on the other, machine learning algorithms need to be insensitive to outliers while being able to model the context. Whereas the latter has been tackled by the use of Long Short-Term Memory (LSTM) networks, the former is still under very active investigations, even though more than a decade of research has provided a large set of acoustic descriptors. In this paper, we propose a solution to the problem of ‘context-aware’ emotional relevant feature extraction, by combining Convolutional Neural Networks (CNNs) with LSTM networks, in order to automatically learn the best representation of the speech signal directly from the raw time representation. In this novel work on the so-called end-to-end speech emotion recognition, we show that the use of the proposed topology significantly outperforms the traditional approaches based on signal processing techniques for the prediction of spontaneous and natural emotions on the RECOLA database.",
"",
"Standard deep neural network-based acoustic models for automatic speech recognition (ASR) rely on hand-engineered input features, typically log-mel filterbank magnitudes. In this paper, we describe a convolutional neural network - deep neural network (CNN-DNN) acoustic model which takes raw multichannel waveforms as input, i.e. without any preceding feature extraction, and learns a similar feature representation through supervised training. By operating directly in the time domain, the network is able to take advantage of the signal's fine time structure that is discarded when computing filterbank magnitude features. This structure is especially useful when analyzing multichannel inputs, where timing differences between input channels can be used to localize a signal in space. The first convolutional layer of the proposed model naturally learns a filterbank that is selective in both frequency and direction of arrival, i.e. a bank of bandpass beamformers with an auditory-like frequency scale. When trained on data corrupted with noise coming from different spatial locations, the network learns to filter them out by steering nulls in the directions corresponding to the noise sources. Experiments on a simulated multichannel dataset show that the proposed acoustic model outperforms a DNN that uses log-mel filterbank magnitude features under noisy and reverberant conditions.",
"Speaker verification systems traditionally extract and model cepstral features or filter bank energies from the speech signal. In this paper, inspired by the success of neural network-based approaches to model directly raw speech signal for applications such as speech recognition, emotion recognition and anti-spoofing, we propose a speaker verification approach where speaker discriminative information is directly learned from the speech signal by: (a) first training a CNN-based speaker identification system that takes as input raw speech signal and learns to classify on speakers (unknown to the speaker verification system); and then (b) building a speaker detector for each speaker in the speaker verification system by replacing the output layer of the speaker identification system by two outputs (genuine, impostor), and adapting the system in a discriminative manner with enrollment speech of the speaker and impostor speech data. Our investigations on the Voxforge database shows that this approach can yield systems competitive to state-of-the-art systems. An analysis of the filters in the first convolution layer shows that the filters give emphasis to information in low frequency regions (below 1000 Hz) and implicitly learn to model fundamental frequency information in the speech signal for speaker discrimination.",
"Deep neural networks (DNN) have achieved significant success in the field of speech recognition. One of the main advantages of the DNN is automatic feature extraction without human intervention. Therefore, we incorporate a pseudo-filterbank layer to the bottom of DNN and train the whole filterbank layer and the following networks jointly, while most systems take pre-defined mel-scale filterbanks as acoustic features to DNN. In the experiment, we use Gaussian functions instead of triangular mel-scale filterbanks. This technique enables a filterbank layer to maintain the functionality of frequency domain smoothing. The proposed method provides an 8.0 relative improvement in clean condition on ASJ+JNAS corpus and a 2.7 relative improvement on noise-corrupted ASJ+JNAS corpus compared with traditional fully-connected DNN. Experimental results show that the frame-level transformation of filterbank layer constrains flexibility and promotes learning efficiency in acoustic modeling.",
"Albeit recent progress in speaker verification generates powerful models, malicious attacks in the form of spoofed speech, are generally not coped with. Recent results in ASVSpoof2015 and BTAS2016 challenges indicate that spoof-aware features are a possible solution to this problem. Most successful methods in both challenges focus on spoof-aware features, rather than focusing on a powerful classifier. In this paper we present a novel raw waveform based deep model for spoofing detection, which jointly acts as a feature extractor and classifier, thus allowing it to directly classify speech signals. This approach can be considered as an end-to-end classifier, which removes the need for any pre- or post-processing on the data, making training and evaluation a streamlined process, consuming less time than other neural-network based approaches. The experiments on the BTAS2016 dataset show that the system performance is significantly improved by the proposed raw waveform convolutional long short term neural network (CLDNN), from the previous best published 1.26 half total error rate (HTER) to the current 0.82 HTER. Moreover it shows that the proposed system also performs well under the unknown (RE-PH2-PH3,RE-LPPH2-PH3) conditions.",
"",
"The effectiveness of introducing deep neural networks into conventional speaker recognition pipelines has been broadly shown to benefit system performance. A novel text-independent speaker verification (SV) framework based on the triplet loss and a very deep convolutional neural network architecture (i.e., Inception-Resnet-v1) are investigated in this study, where a fixed-length speaker discriminative embedding is learned from sparse speech features and utilized as a feature representation for the SV tasks. A concise description of the neural network based speaker discriminative training with triplet loss is presented. An Euclidean distance similarity metric is applied in both network training and SV testing, which ensures the SV system to follow an end-to-end fashion. By replacing the final max average pooling layer with a spatial pyramid pooling layer in the Inception-Resnet-v1 architecture, the fixed-length input constraint is relaxed and an obvious performance gain is achieved compared with the fixed-length input speaker embedding system. For datasets with more severe training test condition mismatches, the probabilistic linear discriminant analysis (PLDA) back end is further introduced to replace the distance based scoring for the proposed speaker embedding system. Thus, we reconstruct the SV task with a neural network based front-end speaker embedding system and a PLDA that provides channel and noise variabilities compensation in the back end. Extensive experiments are conducted to provide useful hints that lead to a better testing performance. Comparison with the state-of-the-art SV frameworks on three public datasets (i.e., a prompt speech corpus, a conversational speech Switchboard corpus, and NIST SRE10 10 s–10 s condition) justifies the effectiveness of our proposed speaker embedding system.",
"Abstract Automaticspeechrecognitionsystemstypicallymodeltherela-tionship between the acoustic speech signal and the phones intwo separate steps: feature extraction and classier training. Inourrecentworks, wehaveshownthat, intheframeworkofcon-volutionalneuralnetworks(CNN),therelationshipbetweentheraw speech signal and the phones can be directly modeled andASR systems competitive to standard approach can be built. Inthis paper, we rst analyze and show that, between the rst twoconvolutional layers, the CNN learns (in parts) and models thephone-specic spectral envelope information of 2-4 ms speech.Given that we show that the CNN-based approach yields ASRtrends similar to standard short-term spectral based ASR sys-tem under mismatched (noisy) conditions, with the CNN-basedapproach being more robust.Index Terms: automatic speech recognition, convolutionalneural networks, raw signal, robust speech recognition. 1. Introduction State-of-the-art automatic speech recognition (ASR) systemstypically model the relationship between the acoustic speechsignal and the phones in two separate steps, which are op-timized in an independent manner [1]. In a rst step, thespeech signal is transformed into features, usually composed ofa dimensionality reduction phase and an information selectionphase, based on the task-specic knowledge of the phenomena.These two phases have been carefully hand-crafted, leading tostate-of-the-art features such as Mel frequency cepstral coef-cients(MFCCs)orperceptuallinearpredictioncepstralfeatures(PLPs). In a second step, the likelihood of subword units suchas, phonemes is estimated using generative models or discrimi-native models.In recent years, in the hybrid HMM ANN framework [1],there has been growing interests in using intermediate rep-resentations instead of conventional features, such as cepstral-based features, as input for neural networks-based systems.ANNs with deep learning architectures, more precisely, deepneural networks (DNNs) [2, 3], which can yield better systemthan a single hidden layer MLP have been proposed to addressvarious aspects of acoustic modeling. More specically, useof context-dependent phonemes [4, 5]; use of spectral featuresas opposed to cepstral features [6, 7]; CNN-based system withMel lter bank energies as input [8, 9, 10]; combination of dif-ferent features [11], to name a few. Features learning from therawspeechsignalusingneuralnetworks-basedsystemshasalsobeen investigated in [12]. In all these approaches, the features",
"",
"Mel-filter banks are commonly used in speech recognition, as they are motivated from theory related to speech production and perception. While features derived from mel-filter banks are quite popular, we argue that this filter bank is not really an appropriate choice as it is not learned for the objective at hand, i.e. speech recognition. In this paper, we explore replacing the filter bank with a filter bank layer that is learned jointly with the rest of a deep neural network. Thus, the filter bank is learned to minimize cross-entropy, which is more closely tied to the speech recognition objective. On a 50-hour English Broadcast News task, we show that we can achieve a 5 relative improvement in word error rate (WER) using the filter bank learning approach, compared to having a fixed set of filters.",
"In this paper we propose a novel model for unconditional audio generation task that generates one audio sample at a time. We show that our model which profits from combining memory-less modules, namely autoregressive multilayer perceptron, and stateful recurrent neural networks in a hierarchical structure is de facto powerful to capture the underlying sources of variations in temporal domain for very long time on three datasets of different nature. Human evaluation on the generated samples indicate that our model is preferred over competing models. We also show how each component of the model contributes to the exhibited performance.",
"With the development of speech synthesis techniques, automatic speaker verification systems face the serious challenge of spoofing attack. In order to improve the reliability of speaker verification systems, we develop a new filter bank-based cepstral feature, deep neural network (DNN) filter bank cepstral coefficients, to distinguish between natural and spoofed speech. The DNN filter bank is automatically generated by training a filter bank neural network (FBNN) using natural and synthetic speech. By adding restrictions on the training rules, the learned weight matrix of FBNN is band limited and sorted by frequency, similar to the normal filter bank. Unlike the manually designed filter bank, the learned filter bank has different filter shapes in different channels, which can capture the differences between natural and synthetic speech more effectively. The experimental results on the ASVspoof 2015 database show that the Gaussian mixture model maximum-likelihood classifier trained by the new feature performs better than the state-of-the-art linear frequency triangle filter bank cepstral coefficients-based classifier, especially on detecting unknown attacks.",
"",
"",
"Learning an acoustic model directly from the raw waveform has been an active area of research. However, waveformbased models have not yet matched the performance of logmel trained neural networks. We will show that raw waveform features match the performance of log-mel filterbank energies when used with a state-of-the-art CLDNN acoustic model trained on over 2,000 hours of speech. Specifically, we will show the benefit of the CLDNN, namely the time convolution layer in reducing temporal variations, the frequency convolution layer for preserving locality and reducing frequency variations, as well as the LSTM layers for temporal modeling. In addition, by stacking raw waveform features with log-mel features, we achieve a 3 relative reduction in word error rate."
]
}
|
1811.09725
|
2901616798
|
Deep learning is currently playing a crucial role toward higher levels of artificial intelligence. This paradigm allows neural networks to learn complex and abstract representations, that are progressively obtained by combining simpler ones. Nevertheless, the internal "black-box" representations automatically discovered by current neural architectures often suffer from a lack of interpretability, making of primary interest the study of explainable machine learning techniques. This paper summarizes our recent efforts to develop a more interpretable neural model for directly processing speech from the raw waveform. In particular, we propose SincNet, a novel Convolutional Neural Network (CNN) that encourages the first layer to discover more meaningful filters by exploiting parametrized sinc functions. In contrast to standard CNNs, which learn all the elements of each filter, only low and high cutoff frequencies of band-pass filters are directly learned from data. This inductive bias offers a very compact way to derive a customized filter-bank front-end, that only depends on some parameters with a clear physical meaning. Our experiments, conducted on both speaker and speech recognition, show that the proposed architecture converges faster, performs better, and is more interpretable than standard CNNs.
|
Similar to SincNet, some previous works have proposed to add constraints on the CNN filters, for instance forcing them to work on specific bands @cite_59 @cite_47 . Differently from the proposed approach, the latter works operate on spectrogram features and still learn all the L elements of the CNN filters. An idea related to the proposed method has been recently explored in @cite_28 , where a set of parameterized Gaussian filters are employed. This approach operates on the spectrogram domain, while SincNet directly considers the raw waveform in the time domain.
|
{
"cite_N": [
"@cite_28",
"@cite_47",
"@cite_59"
],
"mid": [
"2658929981",
"2587891104",
"1969851134"
],
"abstract": [
"Deep neural networks (DNN) have achieved significant success in the field of speech recognition. One of the main advantages of the DNN is automatic feature extraction without human intervention. Therefore, we incorporate a pseudo-filterbank layer to the bottom of DNN and train the whole filterbank layer and the following networks jointly, while most systems take pre-defined mel-scale filterbanks as acoustic features to DNN. In the experiment, we use Gaussian functions instead of triangular mel-scale filterbanks. This technique enables a filterbank layer to maintain the functionality of frequency domain smoothing. The proposed method provides an 8.0 relative improvement in clean condition on ASJ+JNAS corpus and a 2.7 relative improvement on noise-corrupted ASJ+JNAS corpus compared with traditional fully-connected DNN. Experimental results show that the frame-level transformation of filterbank layer constrains flexibility and promotes learning efficiency in acoustic modeling.",
"With the development of speech synthesis techniques, automatic speaker verification systems face the serious challenge of spoofing attack. In order to improve the reliability of speaker verification systems, we develop a new filter bank-based cepstral feature, deep neural network (DNN) filter bank cepstral coefficients, to distinguish between natural and spoofed speech. The DNN filter bank is automatically generated by training a filter bank neural network (FBNN) using natural and synthetic speech. By adding restrictions on the training rules, the learned weight matrix of FBNN is band limited and sorted by frequency, similar to the normal filter bank. Unlike the manually designed filter bank, the learned filter bank has different filter shapes in different channels, which can capture the differences between natural and synthetic speech more effectively. The experimental results on the ASVspoof 2015 database show that the Gaussian mixture model maximum-likelihood classifier trained by the new feature performs better than the state-of-the-art linear frequency triangle filter bank cepstral coefficients-based classifier, especially on detecting unknown attacks.",
"Mel-filter banks are commonly used in speech recognition, as they are motivated from theory related to speech production and perception. While features derived from mel-filter banks are quite popular, we argue that this filter bank is not really an appropriate choice as it is not learned for the objective at hand, i.e. speech recognition. In this paper, we explore replacing the filter bank with a filter bank layer that is learned jointly with the rest of a deep neural network. Thus, the filter bank is learned to minimize cross-entropy, which is more closely tied to the speech recognition objective. On a 50-hour English Broadcast News task, we show that we can achieve a 5 relative improvement in word error rate (WER) using the filter bank learning approach, compared to having a fixed set of filters."
]
}
|
1811.09725
|
2901616798
|
Deep learning is currently playing a crucial role toward higher levels of artificial intelligence. This paradigm allows neural networks to learn complex and abstract representations, that are progressively obtained by combining simpler ones. Nevertheless, the internal "black-box" representations automatically discovered by current neural architectures often suffer from a lack of interpretability, making of primary interest the study of explainable machine learning techniques. This paper summarizes our recent efforts to develop a more interpretable neural model for directly processing speech from the raw waveform. In particular, we propose SincNet, a novel Convolutional Neural Network (CNN) that encourages the first layer to discover more meaningful filters by exploiting parametrized sinc functions. In contrast to standard CNNs, which learn all the elements of each filter, only low and high cutoff frequencies of band-pass filters are directly learned from data. This inductive bias offers a very compact way to derive a customized filter-bank front-end, that only depends on some parameters with a clear physical meaning. Our experiments, conducted on both speaker and speech recognition, show that the proposed architecture converges faster, performs better, and is more interpretable than standard CNNs.
|
This paper extends our previous studies on the SincNet @cite_6 . To the best of our knowledge, this paper is the first that shows the effectiveness of the proposed SincNet in a speech recognition application. Moreover, this work not only considers standard close-talking speech recognition, but it also extends the validation of SincNet to distant-talking speech recognition @cite_0 @cite_61 @cite_56 .
|
{
"cite_N": [
"@cite_0",
"@cite_56",
"@cite_6",
"@cite_61"
],
"mid": [
"2779818703",
"2964067718",
"2964052309",
"2962816167"
],
"abstract": [
"Deep learning is an emerging technology that is considered one of the most promising directions for reaching higher levels of artificial intelligence. Among the other achievements, building computers that understand speech represents a crucial leap towards intelligent machines. Despite the great efforts of the past decades, however, a natural and robust human-machine speech interaction still appears to be out of reach, especially when users interact with a distant microphone in noisy and reverberant environments. The latter disturbances severely hamper the intelligibility of a speech signal, making Distant Speech Recognition (DSR) one of the major open challenges in the field. This thesis addresses the latter scenario and proposes some novel techniques, architectures, and algorithms to improve the robustness of distant-talking acoustic models. We first elaborate on methodologies for realistic data contamination, with a particular emphasis on DNN training with simulated data. We then investigate on approaches for better exploiting speech contexts, proposing some original methodologies for both feed-forward and recurrent neural networks. Lastly, inspired by the idea that cooperation across different DNNs could be the key for counteracting the harmful effects of noise and reverberation, we propose a novel deep learning paradigm called “network of deep neural networks”. The analysis of the original concepts were based on extensive experimental validations conducted on both real and simulated data, considering different corpora, microphone configurations, environments, noisy conditions, and ASR tasks.",
"",
"Deep learning is progressively gaining popularity as a viable alternative to i-vectors for speaker recognition. Promising results have been recently obtained with Convolutional Neural Networks (CNNs) when fed by raw speech samples directly. Rather than employing standard hand-crafted features, the latter CNNs learn low-level speech representations from waveforms, potentially allowing the network to better capture important narrow-band speaker characteristics such as pitch and formants. Proper design of the neural network is crucial to achieve this goal.This paper proposes a novel CNN architecture, called SincNet, that encourages the first convolutional layer to discover more meaningful filters. SincNet is based on parametrized sinc functions, which implement band-pass filters. In contrast to standard CNNs, that learn all elements of each filter, only low and high cutoff frequencies are directly learned from data with the proposed method. This offers a very compact and efficient way to derive a customized filter bank specifically tuned for the desired application.Our experiments, conducted on both speaker identification and speaker verification tasks, show that the proposed architecture converges faster and performs better than a standard CNN on raw waveforms.",
"Despite the remarkable progress recently made in distant speech recognition, state-of-the-art technology still suffers from a lack of robustness, especially when adverse acoustic conditions characterized by non-stationary noises and reverberation are met."
]
}
|
1906.11229
|
2954295542
|
Performance and scalability are a major concern for blockchain systems to become viable for mainstream applications. While many permissionless systems are limited by slow the consensus algorithms, Hyperledger Fabric has unique throughput optimization potential due to its permissioned nature. It has been shown to handle tens of thousands of transactions per second. However, these numbers show only the nominal throughput for uncontested transaction workloads. If incoming transactions compete for a small set of hot keys in the world state, the effective throughput drops drastically. We propose a novel two-pronged transaction execution approach that minimizes invalid transactions in the Fabric blockchain while maximizing concurrent execution.
|
We base this work on , our previous optimization of Hyperledger Fabric @cite_7 . We introduced more efficient data structures, caching and increased parallelization in the transaction validation pipeline to increase Fabric's throughput for conflict-free workloads by a factor six to seven. The numbers we presented resulted from a conflict-free transaction workload. Now, we extend our findings to handle arbitrary contentious workloads.
|
{
"cite_N": [
"@cite_7"
],
"mid": [
"2964101012"
],
"abstract": [
"Blockchain technologies are expected to make a significant impact on a variety of industries. However, one issue holding them back is their limited transaction throughput, especially compared to established solutions such as distributed database systems. In this paper, we re-architect a modern permissioned blockchain system, Hyperledger Fabric, to increase transaction throughput from 3,000 to 20,000 transactions per second. We focus on performance bottlenecks beyond the consensus mechanism, and we propose architectural changes that reduce computation and I O overhead during transaction ordering and validation to greatly improve throughput. Notably, our optimizations are fully plug-and-play and do not require any interface changes to Hyperledger Fabric."
]
}
|
1906.11229
|
2954295542
|
Performance and scalability are a major concern for blockchain systems to become viable for mainstream applications. While many permissionless systems are limited by slow the consensus algorithms, Hyperledger Fabric has unique throughput optimization potential due to its permissioned nature. It has been shown to handle tens of thousands of transactions per second. However, these numbers show only the nominal throughput for uncontested transaction workloads. If incoming transactions compete for a small set of hot keys in the world state, the effective throughput drops drastically. We propose a novel two-pronged transaction execution approach that minimizes invalid transactions in the Fabric blockchain while maximizing concurrent execution.
|
Amiri @cite_4 introduce their ParBlockchain using a very similar architecture to Fabric's but with an OX model. Here, the ordering service also generates a dependency graph of the transactions an a block. Subsequently, all transactions in the new block are distributed to nodes in the network to be executed, taking the dependencies into account. Only a subset of nodes executes any given transaction and shares the result with the rest of the network. Their approach has two major drawbacks. First, they require the ordering service to figure out transaction dependencies before they are executed. Not only would the orderers have to have complete knowledge about all installed smart contracts to do that, it would also drastically restrict the complexity of allowed contracts. Even if just a single conditional statement relies on a state value, for example , reasoning about the result becomes impossible. Second, depending on the workload it can be necessary that all nodes have to communicate the current world state after every transaction execution to resolve execution deadlocks. This leads to a vast networking overhead.
|
{
"cite_N": [
"@cite_4"
],
"mid": [
"2914269357"
],
"abstract": [
"Many existing blockchains do not adequately address all the characteristics of distributed system applications and suffer from serious architectural limitations resulting in performance and confidentiality issues. While recent permissioned blockchain systems, have tried to overcome these limitations, their focus has mainly been on workloads with no-contention, i.e., no conflicting transactions. In this paper, we introduce OXII, a new paradigm for permissioned blockchains to support distributed applications that execute concurrently. OXII is designed for workloads with (different degrees of) contention. We then present ParBlockchain, a permissioned blockchain designed specifically in the OXII paradigm. The evaluation of ParBlockchain using a series of benchmarks reveals that its performance in workloads with any degree of contention is better than the state of the art permissioned blockchain systems."
]
}
|
1906.11229
|
2954295542
|
Performance and scalability are a major concern for blockchain systems to become viable for mainstream applications. While many permissionless systems are limited by slow the consensus algorithms, Hyperledger Fabric has unique throughput optimization potential due to its permissioned nature. It has been shown to handle tens of thousands of transactions per second. However, these numbers show only the nominal throughput for uncontested transaction workloads. If incoming transactions compete for a small set of hot keys in the world state, the effective throughput drops drastically. We propose a novel two-pronged transaction execution approach that minimizes invalid transactions in the Fabric blockchain while maximizing concurrent execution.
|
Sharma @cite_0 approach blockchains from a classical database point of view and attempt to incorporate concepts like early abort and transaction reordering into Hyperledger Fabric. However, they ignore its modular design and closely couple the different building blocks. For both early abort and transaction reordering the ordering service needs to have a deep understanding of the transaction content to be able to unpack and analyze RW sets. Furthermore, transaction reordering only works in pathological cases. Whenever a key appears both in the read and write set, which is the case for any application that transfers any kind of asset, reordering will not eliminate RW set conflicts. While their early transaction abort might increase overall throughput slightly, it cannot solve the problem of hot keys and only skews the transaction workload away from those keys.
|
{
"cite_N": [
"@cite_0"
],
"mid": [
"2898826023"
],
"abstract": [
"Within the last few years, a countless number of blockchain systems have emerged on the market, each one claiming to revolutionize the way of distributed transaction processing in one way or the other. Many blockchain features, such as byzantine fault tolerance (BFT), are indeed valuable additions in modern environments. However, despite all the hype around the technology, many of the challenges that blockchain systems have to face are fundamental transaction management problems. These are largely shared with traditional database systems, which have been around for decades already. These similarities become especially visible for systems, that blur the lines between blockchain systems and classical database systems. A great example of this is Hyperledger Fabric, an open-source permissioned blockchain system under development by IBM. By having a relaxed view on BFT, the transaction pipeline of Fabric highly resembles the workflow of classical distributed databases systems. This raises two questions: (1) Which conceptual similarities and differences do actually exist between a system such as Fabric and a classical distributed database system? (2) Is it possible to improve on the performance of Fabric by transitioning technology from the database world to blockchains and thus blurring the lines between these two types of systems even further? To tackle these questions, we first explore Fabric from the perspective of database research, where we observe weaknesses in the transaction pipeline. We then solve these issues by transitioning well-understood database concepts to Fabric, namely transaction reordering as well as early transaction abort. Our experimental evaluation shows that our improved version Fabric++ significantly increases the throughput of successful transactions over the vanilla version by up to a factor of 3x."
]
}
|
1906.11229
|
2954295542
|
Performance and scalability are a major concern for blockchain systems to become viable for mainstream applications. While many permissionless systems are limited by slow the consensus algorithms, Hyperledger Fabric has unique throughput optimization potential due to its permissioned nature. It has been shown to handle tens of thousands of transactions per second. However, these numbers show only the nominal throughput for uncontested transaction workloads. If incoming transactions compete for a small set of hot keys in the world state, the effective throughput drops drastically. We propose a novel two-pronged transaction execution approach that minimizes invalid transactions in the Fabric blockchain while maximizing concurrent execution.
|
Lastly, Zhang @cite_3 present a solution for a client-side early abort mechanism for Fabric. They introduce a transaction cache on the client that analyzes endorsed transactions to detect RW set conflicts and only sends conflict-free transactions to the ordering service. Transactions that have dependencies are held in the cache until the conflict is resolved and then they are sent back to the endorsers for re-execution. This approach prevents invalid transactions from a single client, but cannot deal with conflicts between multiple clients. Moreover, it cannot deal with hot key workloads either.
|
{
"cite_N": [
"@cite_3"
],
"mid": [
"2954107585"
],
"abstract": [
"Blockchain is one of the most popular distributed ledger technologies. In order to solve the trustless problems between enterprises, many permissioned blockchain platforms are proposed in recent years. Hyperledger Fabric is a permissioned blockchain which is comprised of chaincode, endorsing peers, ordering service, committing peers and membership service. Fabric provides identity authentication, transactions encryption and high performance. Hence, decentralized application based on Fabric can be applied among enterprises to implement the immutable and high throughput for transactions. However, some limitations still exist in Fabric framework. This paper surveys the risk of non-deterministic transactions in Fabric, which can be caused by read-write conflict and transactions order dependency. Transactions are judged to be invalid by committing peers when read-write conflict exists. Besides, transactions submission order may be disrupted during endorsing and ordering. Both of the two risks can make the final transaction execution result non-deterministic. Finally, a solution of cache layer on Fabric client is proposed to solve the risk of non-deterministic transactions."
]
}
|
1906.11211
|
2955224228
|
Encouraged by the success of deep learning in a variety of domains, we investigate the suitability and effectiveness of Recurrent Neural Networks (RNNs) in a domain where deep learning has not yet been used; namely detecting confusion from eye-tracking data. Through experiments with a dataset of user interactions with ValueChart (an interactive visualization tool), we found that RNNs learn a feature representation from the raw data that allows for a more powerful classifier than previous methods that use engineered features. This is evidenced by the stronger performance of the RNN (0.74 0.71 sensitivity specificity), as compared to a Random Forest classifier (0.51 0.70 sensitivity specificity), when both are trained on an un-augmented dataset. However, using engineered features allows for simple data augmentation methods to be used. These same methods are not as effective at augmentation for the feature representation learned from the raw data, likely due to an inability to match the temporal dynamics of the data.
|
Most work on predicting affect with deep learning has been in computer vision and natural language, where established methods already exist for classifying emotions from pictures, video, and sound @cite_9 @cite_18 @cite_19 . These works focus on predicting one of anger, disgust, fear, happy, sad, surprise, or neutral. There is less work that uses DL for emotion recognition based on data from live interaction with users, which is ultimately what we want from an affect-sensitive artificial agent.
|
{
"cite_N": [
"@cite_19",
"@cite_9",
"@cite_18"
],
"mid": [
"",
"2277498883",
"2414501075"
],
"abstract": [
"",
"Deep learning based approaches to facial analysis and video analysis have recently demonstrated high performance on a variety of key tasks such as face recognition, emotion recognition and activity recognition. In the case of video, information often must be aggregated across a variable length sequence of frames to produce a classification result. Prior work using convolutional neural networks (CNNs) for emotion recognition in video has relied on temporal averaging and pooling operations reminiscent of widely used approaches for the spatial aggregation of information. Recurrent neural networks (RNNs) have seen an explosion of recent interest as they yield state-of-the-art performance on a variety of sequence analysis tasks. RNNs provide an attractive framework for propagating information over a sequence using a continuous valued hidden layer representation. In this work we present a complete system for the 2015 Emotion Recognition in the Wild (EmotiW) Challenge. We focus our presentation and experimental analysis on a hybrid CNN-RNN architecture for facial expression analysis that can outperform a previously applied CNN approach using temporal averaging for aggregation.",
"Despite growing research interest, emotion understanding for user-generated videos remains a challenging problem. Major obstacles include the diversity and complexity of video content, as well as the sparsity of expressed emotions. For the first time, we systematically study large-scale video emotion recognition by transferring deep feature encodings. In addition to the traditional, supervised recognition, we study the problem of zero-shot emotion recognition, where emotions in the test set are unseen during training. To cope with this task, we utilize knowledge transferred from auxiliary image and text corpora. A novel auxiliary Image Transfer Encoding (ITE) process is proposed to efficiently encode and generate video representation. We also thoroughly investigate different configurations of convolutional neural networks. Comprehensive experiments on multiple datasets demonstrate the effectiveness of our framework."
]
}
|
1906.11211
|
2955224228
|
Encouraged by the success of deep learning in a variety of domains, we investigate the suitability and effectiveness of Recurrent Neural Networks (RNNs) in a domain where deep learning has not yet been used; namely detecting confusion from eye-tracking data. Through experiments with a dataset of user interactions with ValueChart (an interactive visualization tool), we found that RNNs learn a feature representation from the raw data that allows for a more powerful classifier than previous methods that use engineered features. This is evidenced by the stronger performance of the RNN (0.74 0.71 sensitivity specificity), as compared to a Random Forest classifier (0.51 0.70 sensitivity specificity), when both are trained on an un-augmented dataset. However, using engineered features allows for simple data augmentation methods to be used. These same methods are not as effective at augmentation for the feature representation learned from the raw data, likely due to an inability to match the temporal dynamics of the data.
|
Work that uses deep learning for classifying confusion has applied RNNs to sequential interaction data @cite_16 @cite_12 . These works seek to predict multiple emotions (one of which is confusion) while students interact with an Intelligent Tutoring System. While these works utilize interaction data, we utilize eye-tracking data, which has been shown to be a good predictor of emotional or attentional states such as mind wandering @cite_14 , as well as boredom, and curiosity, while learning with educational software @cite_8 . @cite_22 predicted confusion using eye-tracking data, and achieved state of the art results using the Random Forest (RF) algorithm (Section 4). In @cite_13 , a patient's gaze point was superimposed onto the face of a doctor from the patient's point of view. The result is a video that shows the patient's gaze point over time. An RNN was then used to predict a developmental disorder. To the best of our knowledge, deep learning has yet to be used for any affect predictive task based on eye-tracking data.
|
{
"cite_N": [
"@cite_14",
"@cite_22",
"@cite_8",
"@cite_16",
"@cite_13",
"@cite_12"
],
"mid": [
"2111001556",
"2507410144",
"107670845",
"2809165304",
"2523868740",
"2699539148"
],
"abstract": [
"Mind wandering (MW) is a ubiquitous phenomenon where attention involuntarily shifts from task-related processing to task-unrelated thoughts. There is a need for adaptive systems that can reorient attention when MW is detected due to its detrimental effects on performance and productivity. This paper proposes an automated gaze-based detector of self-caught MW (i.e., when users become consciously aware that they are MW). Eye gaze data and self-reports of MW were collected as 178 users read four instructional texts from a computer interface. Supervised machine learning models trained on features extracted from users’ gaze fixations were used to detect pages where users caught themselves MW. The best performing model achieved a user-independent kappa of .45 (accuracy of 74 compared to a chance accuracy of 52 ); the first ever demonstration of a self-caught MW detector. An analysis of the features revealed that during MW, users made more regression fixations, had longer saccades that crossed lines more often, and had more uniform fixation durations, indicating a violation from normal reading patterns. Applications of the MW detector are discussed.",
"Confident usage of information visualizations is thought to be influenced by cognitive aspects as well as amount of exposure and training. To support the development of individual competency in visualization processing, it is important to ascertain if we can track users' progress or difficulties they might have while working with a given visualization. In this paper, we extend previous work on predicting in real time a user's learning curve--a mathematical model that can represent a user's skill acquisition ability--when working with a visualization. First, we investigate whether results we previously obtained in predicting users' learning curves during visualization processing generalize to a different visualization. Second, we study to what extent we can make predictions on a user's learning curve without information on the visualization being used. Our models leverage various data sources, including a user's gaze behavior, pupil dilation, and cognitive abilities. We show that these models outperform a baseline that leverages knowledge on user task performance so far. Our best performing model achieves good accuracies in predicting users' learning curves even after observing users' performance on a few tasks only. These results represent an important step toward understanding how to support users in learning a new visualization.",
"In this paper we investigate the usefulness of eye tracking data for predicting emotions relevant to learning, specifically boredom and curiosity. The data was collected during a study with MetaTutor, an intelligent tutoring system ITS designed to promote the use of self-regulated learning strategies. We used a variety of machine learning and feature selection techniques to predict students' self-reported emotions from gaze data features. We examined the optimal amount of interaction time needed to make predictions, as well as which features are most predictive of each emotion. The findings provide insight into how to detect when students disengage from MetaTutor.",
"The past few years have seen a surge of interest in deep neural networks. The wide application of deep learning in other domains such as image classification has driven considerable recent interest and efforts in applying these methods in educational domains. However, there is still limited research comparing the predictive power of the deep learning approach with the traditional feature engineering approach for common student modeling problems such as sensor-free affect detection. This paper aims to address this gap by presenting a thorough comparison of several deep neural network approaches with a traditional feature engineering approach in the context of affect and behavior modeling. We built detectors of student affective states and behaviors as middle school students learned science in an open-ended learning environment called Betty’s Brain, using both approaches. Overall, we observed a tradeoff where the feature engineering models were better when considering a single optimized threshold (for intervention), whereas the deep learning models were better when taking model confidence fully into account (for discovery with models analyses).",
"This paper proposes a system for fine-grained classification of developmental disorders via measurements of individuals’ eye-movements using multi-modal visual data. While the system is engineered to solve a psychiatric problem, we believe the underlying principles and general methodology will be of interest not only to psychiatrists but to researchers and engineers in medical machine vision. The idea is to build features from different visual sources that capture information not contained in either modality. Using an eye-tracker and a camera in a setup involving two individuals speaking, we build temporal attention features that describe the semantic location that one person is focused on relative to the other person’s face. In our clinical context, these temporal attention features describe a patient’s gaze on finely discretized regions of an interviewing clinician’s face, and are used to classify their particular developmental disorder.",
"Affect detection has become a prominent area in student modeling in the last decade and considerable progress has been made in developing effective models. Many of the most successful models have leveraged physical and physiological sensors to accomplish this. While successful, such systems are difficult to deploy at scale due to economic and political constraints, limiting the utility of their application. Examples of “sensor-free” affect detectors that assess students based solely using data on the interaction between students and computer-based learning platforms exist, but these detectors generally have not reached high enough levels of quality to justify their use in real-time interventions. However, the classification algorithms used in these previous sensor-free detectors have not taken full advantage of the newest methods emerging in the field. The use of deep learning algorithms, such as recurrent neural networks (RNNs), have been applied to a range of other domains including pattern recognition and natural language processing with success, but have only recently been attempted in educational contexts. In this work, we construct new “deep” sensor-free affect detectors and report significant improvements over previously reported models."
]
}
|
1906.11024
|
2955646770
|
Recently, the Transformer machine translation system has shown strong results by stacking attention layers on both the source and target-language sides. But the inference of this model is slow due to the heavy use of dot-product attention in auto-regressive decoding. In this paper we speed up Transformer via a fast and lightweight attention model. More specifically, we share attention weights in adjacent layers and enable the efficient re-use of hidden states in a vertical manner. Moreover, the sharing policy can be jointly learned with the MT model. We test our approach on ten WMT and NIST OpenMT tasks. Experimental results show that it yields an average of 1.3X speed-up (with almost no decrease in BLEU) on top of a state-of-the-art implementation that has already adopted a cache for fast inference. Also, our approach obtains a 1.8X speed-up when it works with the model. This is even 16 times faster than the baseline with no use of the attention cache.
|
It has been observed that attention models are critical for state-of-the-art results on many MT tasks @cite_3 @cite_0 @cite_11 . Several research groups have investigated attentive methods for different architectures of neural MT. The earliest is @cite_19 . They introduced an additive attention model into MT systems based on recurrent neural networks (RNNs). More recently, multi-layer attention was successfully applied to convolutional neural MT systems @cite_14 and Transformer systems @cite_11 . In particular, Transformer is popular due to its scalability on large-scale training and the good design of the architecture for implementation.
|
{
"cite_N": [
"@cite_14",
"@cite_3",
"@cite_0",
"@cite_19",
"@cite_11"
],
"mid": [
"2964265128",
"2964308564",
"2525778437",
"2949335953",
"2626778328"
],
"abstract": [
"The prevalent approach to sequence to sequence learning maps an input sequence to a variable length output sequence via recurrent neural networks. We introduce an architecture based entirely on convolutional neural networks. Compared to recurrent models, computations over all elements can be fully parallelized during training to better exploit the GPU hardware and optimization is easier since the number of non-linearities is fixed and independent of the input length. Our use of gated linear units eases gradient propagation and we equip each decoder layer with a separate attention module. We outperform the accuracy of the deep LSTM setup of (2016) on both WMT'14 English-German and WMT'14 English-French translation at an order of magnitude faster speed, both on GPU and CPU.*",
"Abstract: Neural machine translation is a recently proposed approach to machine translation. Unlike the traditional statistical machine translation, the neural machine translation aims at building a single neural network that can be jointly tuned to maximize the translation performance. The models proposed recently for neural machine translation often belong to a family of encoder-decoders and consists of an encoder that encodes a source sentence into a fixed-length vector from which a decoder generates a translation. In this paper, we conjecture that the use of a fixed-length vector is a bottleneck in improving the performance of this basic encoder-decoder architecture, and propose to extend this by allowing a model to automatically (soft-)search for parts of a source sentence that are relevant to predicting a target word, without having to form these parts as a hard segment explicitly. With this new approach, we achieve a translation performance comparable to the existing state-of-the-art phrase-based system on the task of English-to-French translation. Furthermore, qualitative analysis reveals that the (soft-)alignments found by the model agree well with our intuition.",
"Neural Machine Translation (NMT) is an end-to-end learning approach for automated translation, with the potential to overcome many of the weaknesses of conventional phrase-based translation systems. Unfortunately, NMT systems are known to be computationally expensive both in training and in translation inference. Also, most NMT systems have difficulty with rare words. These issues have hindered NMT's use in practical deployments and services, where both accuracy and speed are essential. In this work, we present GNMT, Google's Neural Machine Translation system, which attempts to address many of these issues. Our model consists of a deep LSTM network with 8 encoder and 8 decoder layers using attention and residual connections. To improve parallelism and therefore decrease training time, our attention mechanism connects the bottom layer of the decoder to the top layer of the encoder. To accelerate the final translation speed, we employ low-precision arithmetic during inference computations. To improve handling of rare words, we divide words into a limited set of common sub-word units (\"wordpieces\") for both input and output. This method provides a good balance between the flexibility of \"character\"-delimited models and the efficiency of \"word\"-delimited models, naturally handles translation of rare words, and ultimately improves the overall accuracy of the system. Our beam search technique employs a length-normalization procedure and uses a coverage penalty, which encourages generation of an output sentence that is most likely to cover all the words in the source sentence. On the WMT'14 English-to-French and English-to-German benchmarks, GNMT achieves competitive results to state-of-the-art. Using a human side-by-side evaluation on a set of isolated simple sentences, it reduces translation errors by an average of 60 compared to Google's phrase-based production system.",
"An attentional mechanism has lately been used to improve neural machine translation (NMT) by selectively focusing on parts of the source sentence during translation. However, there has been little work exploring useful architectures for attention-based NMT. This paper examines two simple and effective classes of attentional mechanism: a global approach which always attends to all source words and a local one that only looks at a subset of source words at a time. We demonstrate the effectiveness of both approaches over the WMT translation tasks between English and German in both directions. With local attention, we achieve a significant gain of 5.0 BLEU points over non-attentional systems which already incorporate known techniques such as dropout. Our ensemble model using different attention architectures has established a new state-of-the-art result in the WMT'15 English to German translation task with 25.9 BLEU points, an improvement of 1.0 BLEU points over the existing best system backed by NMT and an n-gram reranker.",
"The dominant sequence transduction models are based on complex recurrent or convolutional neural networks in an encoder-decoder configuration. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 English-to-German translation task, improving over the existing best results, including ensembles by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data."
]
}
|
1906.11052
|
2892456951
|
Data augmentation is a popular technique largely used to enhance the training of convolutional neural networks. Although many of its benefits are well known by deep learning researchers and practitioners, its implicit regularization effects, as compared to popular explicit regularization techniques, such as weight decay and dropout, remain largely unstudied. As a matter of fact, convolutional neural networks for image object classification are typically trained with both data augmentation and explicit regularization, assuming the benefits of all techniques are complementary. In this paper, we systematically analyze these techniques through ablation studies of different network architectures trained with different amounts of training data. Our results unveil a largely ignored advantage of data augmentation: networks trained with just data augmentation more easily adapt to different architectures and amount of training data, as opposed to weight decay and dropout, which require specific fine-tuning of their hyperparameters.
|
Data augmentation was already used in the late 80's and early 90's for handwritten digit recognition @cite_11 and it has been identified as a very important element of many modern successful models, like AlexNet @cite_2 , All-CNN @cite_16 or ResNet @cite_9 , for instance. In some cases, heavy data augmentation has been applied with successful results @cite_17 . In domains other than computer vision, data augmentation has also been proven effective, for example in speech recognition @cite_8 , music source separation @cite_23 or text categorization @cite_4 .
|
{
"cite_N": [
"@cite_11",
"@cite_4",
"@cite_8",
"@cite_9",
"@cite_23",
"@cite_2",
"@cite_16",
"@cite_17"
],
"mid": [
"2111494971",
"2130858665",
"2099621636",
"2194775991",
"2669032454",
"2618530766",
"2963382180",
"1563686443"
],
"abstract": [
"In many machine learning applications, one has access, not only to training data, but also to some high-level a priori knowledge about the desired behavior of the system. For example, it is known in advance that the output of a character recognizer should be invariant with respect to small spatial distortions of the input images (translations, rotations, scale changes, etcetera). We have implemented a scheme that allows a network to learn the derivative of its outputs with respect to distortion operators of our choosing. This not only reduces the learning time and the amount of training data, but also provides a powerful language for specifying what generalizations we wish the network to perform.",
"Abstract Objective Acquiring and representing biomedical knowledge is an increasingly important component of contemporary bioinformatics. A critical step of the process is to identify and retrieve relevant documents among the vast volume of modern biomedical literature efficiently. In the real world, many information retrieval tasks are difficult because of high data dimensionality and the lack of annotated examples to train a retrieval algorithm. Under such a scenario, the performance of information retrieval algorithms is often unsatisfactory, therefore improvements are needed. Design We studied two approaches that enhance the text categorization performance on sparse and high data dimensionality: (1) semantic-preserving dimension reduction by representing text with semantic-enriched features; and (2) augmenting training data with semi-supervised learning. A probabilistic topic model was applied to extract major semantic topics from a corpus of text of interest. The representation of documents was projected from the high-dimensional vocabulary space onto a semantic topic space with reduced dimensionality. A semi-supervised learning algorithm based on graph theory was applied to identify potential positive training cases, which were further used to augment training data. The effects of data transformation and augmentation on text categorization by support vector machine (SVM) were evaluated. Results and Conclusion Semantic-enriched data transformation and the pseudo-positive-cases augmented training data enhance the efficiency and performance of text categorization by SVM.",
"Augmenting datasets by transforming inputs in a way that does not change the label is a crucial ingredient of the state of the art methods for object recognition using neural networks. However this approach has (to our knowledge) not been exploited successfully in speech recognition (with or without neural networks). In this paper we lay the foundation for this approach, and show one way of augmenting speech datasets by transforming spectrograms, using a random linear warping along the frequency dimension. In practice this can be achieved by using warping techniques that are used for vocal tract length normalization (VTLN) - with the difference that a warp factor is generated randomly each time, during training, rather than tting a single warp factor to each training and test speaker (or utterance). At test time, a prediction is made by averaging the predictions over multiple warp factors. When this technique is applied to TIMIT using Deep Neural Networks (DNN) of dierent depths, the Phone Error Rate (PER) improved by an average of 0.65 on the test set. For a Convolutional neural network (CNN) with convolutional layer in the bottom, a gain of 1.0 was observed. These improvements were achieved without increasing the number of training epochs, and suggest that data transformations should be an important component of training neural networks for speech, especially for data limited projects.",
"Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57 error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28 relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.",
"This paper deals with the separation of music into individual instrument tracks which is known to be a challenging problem. We describe two different deep neural network architectures for this task, a feed-forward and a recurrent one, and show that each of them yields themselves state-of-the art results on the SiSEC DSD100 dataset. For the recurrent network, we use data augmentation during training and show that even simple separation networks are prone to overfitting if no data augmentation is used. Furthermore, we propose a blending of both neural network systems where we linearly combine their raw outputs and then perform a multi-channel Wiener filter post-processing. This blending scheme yields the best results that have been reported to-date on the SiSEC DSD100 dataset.",
"We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 different classes. On the test data, we achieved top-1 and top-5 error rates of 37.5 and 17.0 , respectively, which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully connected layers with a final 1000-way softmax. To make training faster, we used non-saturating neurons and a very efficient GPU implementation of the convolution operation. To reduce overfitting in the fully connected layers we employed a recently developed regularization method called \"dropout\" that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3 , compared to 26.2 achieved by the second-best entry.",
"",
"We present a state-of-the-art image recognition system, Deep Image, developed using end-to-end deep learning. The key components are a custom-built supercomputer dedicated to deep learning, a highly optimized parallel algorithm using new strategies for data partitioning and communication, larger deep neural network models, novel data augmentation approaches, and usage of multi-scale high-resolution images. Our method achieves excellent results on multiple challenging computer vision benchmarks."
]
}
|
1906.11052
|
2892456951
|
Data augmentation is a popular technique largely used to enhance the training of convolutional neural networks. Although many of its benefits are well known by deep learning researchers and practitioners, its implicit regularization effects, as compared to popular explicit regularization techniques, such as weight decay and dropout, remain largely unstudied. As a matter of fact, convolutional neural networks for image object classification are typically trained with both data augmentation and explicit regularization, assuming the benefits of all techniques are complementary. In this paper, we systematically analyze these techniques through ablation studies of different network architectures trained with different amounts of training data. Our results unveil a largely ignored advantage of data augmentation: networks trained with just data augmentation more easily adapt to different architectures and amount of training data, as opposed to weight decay and dropout, which require specific fine-tuning of their hyperparameters.
|
Bengio @cite_13 focused on the importance of data augmentation for recognizing handwritten digits through greedy layer-wise unsupervised pre-training @cite_24 . Their main conclusion was that deeper architectures benefit more from data augmentation than shallow networks. Zhang @cite_21 included data augmentation in their analysis of the role of regularization in the generalization of deep networks, although it was considered an explicit regularizer similar to weight decay and dropout. The observation that data augmentation alone outperforms explicitly regularized models for few-shot learning was also made by Hilliard in @cite_19 . Only few works reported the performance of their models when trained with different types of data augmentation levels, as is the case of @cite_10 .
|
{
"cite_N": [
"@cite_10",
"@cite_21",
"@cite_24",
"@cite_19",
"@cite_13"
],
"mid": [
"",
"2566079294",
"2110798204",
"2786398923",
"2137844145"
],
"abstract": [
"",
"Despite their massive size, successful deep artificial neural networks can exhibit a remarkably small difference between training and test performance. Conventional wisdom attributes small generalization error either to properties of the model family, or to the regularization techniques used during training. @PARASPLIT Through extensive systematic experiments, we show how these traditional approaches fail to explain why large neural networks generalize well in practice. Specifically, our experiments establish that state-of-the-art convolutional networks for image classification trained with stochastic gradient methods easily fit a random labeling of the training data. This phenomenon is qualitatively unaffected by explicit regularization, and occurs even if we replace the true images by completely unstructured random noise. We corroborate these experimental findings with a theoretical construction showing that simple depth two neural networks already have perfect finite sample expressivity as soon as the number of parameters exceeds the number of data points as it usually does in practice. @PARASPLIT We interpret our experimental findings by comparison with traditional models.",
"Complexity theory of circuits strongly suggests that deep architectures can be much more efficient (sometimes exponentially) than shallow architectures, in terms of computational elements required to represent some functions. Deep multi-layer neural networks have many levels of non-linearities allowing them to compactly represent highly non-linear and highly-varying functions. However, until recently it was not clear how to train such deep networks, since gradient-based optimization starting from random initialization appears to often get stuck in poor solutions. recently introduced a greedy layer-wise unsupervised learning algorithm for Deep Belief Networks (DBN), a generative model with many layers of hidden causal variables. In the context of the above optimization problem, we study this algorithm empirically and explore variants to better understand its success and extend it to cases where the inputs are continuous or where the structure of the input distribution is not revealing enough about the variable to be predicted in a supervised task. Our experiments also confirm the hypothesis that the greedy layer-wise unsupervised training strategy mostly helps the optimization, by initializing weights in a region near a good local minimum, giving rise to internal distributed representations that are high-level abstractions of the input, bringing better generalization.",
"Learning high quality class representations from few examples is a key problem in metric-learning approaches to few-shot learning. To accomplish this, we introduce a novel architecture where class representations are conditioned for each few-shot trial based on a target image. We also deviate from traditional metric-learning approaches by training a network to perform comparisons between classes rather than relying on a static metric comparison. This allows the network to decide what aspects of each class are important for the comparison at hand. We find that this flexible architecture works well in practice, achieving state-of-the-art performance on the Caltech-UCSD birds fine-grained classification task.",
"Recent theoretical and empirical work in statistical machine learning has demonstrated the potential of learning algorithms for deep architectures, i.e., function classes obtained by composing multiple levels of representation. The hypothesis evaluated here is that intermediate levels of representation, because they can be shared across tasks and examples from different but related distributions, can yield even more benefits. Comparative experiments were performed on a large-scale handwritten character recognition setting with 62 classes (upper case, lower case, digits), using both a multi-task setting and perturbed examples in order to obtain out-ofdistribution examples. The results agree with the hypothesis, and show that a deep learner did beat previously published results and reached human-level performance."
]
}
|
1906.11052
|
2892456951
|
Data augmentation is a popular technique largely used to enhance the training of convolutional neural networks. Although many of its benefits are well known by deep learning researchers and practitioners, its implicit regularization effects, as compared to popular explicit regularization techniques, such as weight decay and dropout, remain largely unstudied. As a matter of fact, convolutional neural networks for image object classification are typically trained with both data augmentation and explicit regularization, assuming the benefits of all techniques are complementary. In this paper, we systematically analyze these techniques through ablation studies of different network architectures trained with different amounts of training data. Our results unveil a largely ignored advantage of data augmentation: networks trained with just data augmentation more easily adapt to different architectures and amount of training data, as opposed to weight decay and dropout, which require specific fine-tuning of their hyperparameters.
|
Recently, the deep learning community seems to have become more aware of the importance of data augmentation. New techniques have been proposed @cite_20 @cite_31 and, very interestingly, models that automatically learn useful data transformations have also been published lately @cite_15 @cite_5 @cite_7 @cite_14 . Another study @cite_30 analyzed the performance of different data augmentation techniques for object recognition and concluded that one of the most successful techniques so far is the traditional transformations carried out in most studies. Finally, a preliminary analysis of the implicit regularization effect of data augmentation was presented in @cite_12 , showing that data augmentation alone provides at least the same generalization performance as weight decay and dropout. The present work follows up on those results and extends the analysis.
|
{
"cite_N": [
"@cite_30",
"@cite_14",
"@cite_7",
"@cite_15",
"@cite_5",
"@cite_31",
"@cite_12",
"@cite_20"
],
"mid": [
"2775795276",
"2770173563",
"2963543962",
"2962920946",
"2604262106",
"2594477595",
"2787919999",
"2746314669"
],
"abstract": [
"In this paper, we explore and compare multiple solutions to the problem of data augmentation in image classification. Previous work has demonstrated the effectiveness of data augmentation through simple techniques, such as cropping, rotating, and flipping input images. We artificially constrain our access to data to a small subset of the ImageNet dataset, and compare each data augmentation technique in turn. One of the more successful data augmentations strategies is the traditional transformations mentioned above. We also experiment with GANs to generate images of different styles. Finally, we propose a method to allow a neural net to learn augmentations that best improve the classifier, which we call neural augmentation. We discuss the successes and shortcomings of this method on various datasets.",
"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 ).",
"Data augmentation is a ubiquitous technique for increasing the size of labeled training sets by leveraging task-specific data transformations that preserve class labels. While it is often easy for domain experts to specify individual transformations, constructing and tuning the more sophisticated compositions typically needed to achieve state-of-the-art results is a time-consuming manual task in practice. We propose a method for automating this process by learning a generative sequence model over user-specified transformation functions using a generative adversarial approach. Our method can make use of arbitrary, non-deterministic transformation functions, is robust to misspecified user input, and is trained on unlabeled data. The learned transformation model can then be used to perform data augmentation for any end discriminative model. In our experiments, we show the efficacy of our approach on both image and text datasets, achieving improvements of 4.0 accuracy points on CIFAR-10, 1.4 F1 points on the ACE relation extraction task, and 3.4 accuracy points when using domain-specific transformation operations on a medical imaging dataset as compared to standard heuristic augmentation approaches.",
"",
"A recurring problem faced when training neural networks is that there is typically not enough data to maximize the generalization capability of deep neural networks. There are many techniques to address this, including data augmentation, dropout, and transfer learning. In this paper, we introduce an additional method, which we call smart augmentation and we show how to use it to increase the accuracy and reduce over fitting on a target network. Smart augmentation works, by creating a network that learns how to generate augmented data during the training process of a target network in a way that reduces that networks loss. This allows us to learn augmentations that minimize the error of that network. Smart augmentation has shown the potential to increase accuracy by demonstrably significant measures on all data sets tested. In addition, it has shown potential to achieve similar or improved performance levels with significantly smaller network sizes in a number of tested cases.",
"Dataset augmentation, the practice of applying a wide array of domain-specific transformations to synthetically expand a training set, is a standard tool in supervised learning. While effective in tasks such as visual recognition, the set of transformations must be carefully designed, implemented, and tested for every new domain, limiting its re-use and generality. In this paper, we adopt a simpler, domain-agnostic approach to dataset augmentation. We start with existing data points and apply simple transformations such as adding noise, interpolating, or extrapolating between them. Our main insight is to perform the transformation not in input space, but in a learned feature space. A re-kindling of interest in unsupervised representation learning makes this technique timely and more effective. It is a simple proposal, but to-date one that has not been tested empirically. Working in the space of context vectors generated by sequence-to-sequence models, we demonstrate a technique that is effective for both static and sequential data.",
"The impressive success of modern deep neural networks on computer vision tasks has been achieved through models of very large capacity compared to the number of available training examples. This overparameterization is often said to be controlled with the help of different regularization techniques, mainly weight decay and dropout. However, since these techniques reduce the effective capacity of the model, typically even deeper and wider architectures are required to compensate for the reduced capacity. Therefore, there seems to be a waste of capacity in this practice. In this paper we build upon recent research that suggests that explicit regularization may not be as important as widely believed and carry out an ablation study that concludes that weight decay and dropout may not be necessary for object recognition if enough data augmentation is introduced.",
"Convolutional neural networks are capable of learning powerful representational spaces, which are necessary for tackling complex learning tasks. However, due to the model capacity required to capture such representations, they are often susceptible to overfitting and therefore require proper regularization in order to generalize well. In this paper, we show that the simple regularization technique of randomly masking out square regions of input during training, which we call cutout, can be used to improve the robustness and overall performance of convolutional neural networks. Not only is this method extremely easy to implement, but we also demonstrate that it can be used in conjunction with existing forms of data augmentation and other regularizers to further improve model performance. We evaluate this method by applying it to current state-of-the-art architectures on the CIFAR-10, CIFAR-100, and SVHN datasets, yielding new state-of-the-art results of 2.56 , 15.20 , and 1.30 test error respectively. Code is available at this https URL"
]
}
|
1906.11061
|
2954210796
|
Data sent over the Internet can be monitored and manipulated by intermediate entities in the data path from the source to the destination. For unencrypted communications (and some encrypted communications with known weaknesses), eavesdropping and man-in-the-middle attacks are possible. For encrypted communication, the identification of the communicating endpoints is still revealed. In addition, encrypted communications may be stored until such time as newly discovered weaknesses in the encryption algorithm or advances in computer hardware render them readable by attackers. In this work, we use public data to evaluate both advertised and observed routes through the Internet and measure the extent to which communications between pairs of countries are exposed to other countries. We use both physical router geolocation as well as the country of registration of the companies owning each router. We find a high level of information exposure; even physically adjacent countries use routes that involve many other countries. We also found that countries that are well connected' tend to be more exposed. Our analysis indicates that there exists a tradeoff between robustness and information exposure in the current Internet.
|
@cite_19 is most similar to our work in that they evaluate routes to determine their information exposure at a country level of abstraction. It uses traceroute and due to limitations of their measurement infrastructure, they limit their analysis to five countries. They focus on measurements of regular users accessing the Alexa Top 100 websites and provide related analytics on specific countries. In contrast, our work is focused on information exposure measurements for the movement of high sensitivity data of interest to foreign nation-states. Instead of being limited to just a few countries, our approach using publicly available datasets can be applied to all countries. Of perhaps greater importance is that our work did not just evaluate router geolocation dataset as in @cite_19 , it includes router country of registration data as well.
|
{
"cite_N": [
"@cite_19"
],
"mid": [
"2404814036"
],
"abstract": [
"An increasing number of countries are passing laws that facilitate the mass surveillance of Internet traffic. In response, governments and citizens are increasingly paying attention to the countries that their Internet traffic traverses. In some cases, countries are taking extreme steps, such as building new Internet Exchange Points (IXPs), which allow networks to interconnect directly, and encouraging local interconnection to keep local traffic local. We find that although many of these efforts are extensive, they are often futile, due to the inherent lack of hosting and route diversity for many popular sites. By measuring the country-level paths to popular domains, we characterize transnational routing detours. We find that traffic is traversing known surveillance states, even when the traffic originates and ends in a country that does not conduct mass surveillance. Then, we investigate how clients can use overlay network relays and the open DNS resolver infrastructure to prevent their traffic from traversing certain jurisdictions. We find that 84 of paths originating in Brazil traverse the United States, but when relays are used for country avoidance, only 37 of Brazilian paths traverse the United States. Using the open DNS resolver infrastructure allows Kenyan clients to avoid the United States on 17 more paths. Unfortunately, we find that some of the more prominent surveillance states (e.g., the U.S.) are also some of the least avoidable countries."
]
}
|
1906.11199
|
2956019919
|
We propose design guidelines for a probabilistic programming facility suitable for deployment as a part of a production software system. As a reference implementation, we introduce Infergo, a probabilistic programming facility for Go, a modern programming language of choice for server-side software development. We argue that a similar probabilistic programming facility can be added to most modern general-purpose programming languages. Probabilistic programming enables automatic tuning of program parameters and algorithmic decision making through probabilistic inference based on the data. To facilitate addition of probabilistic programming capabilities to other programming languages, we share implementation choices and techniques employed in development of Infergo. We illustrate applicability of Infergo to various use cases on case studies, and evaluate Infergo's performance on several benchmarks, comparing Infergo to dedicated inference-centric probabilistic programming frameworks.
|
Automatic differentiation is widely employed in machine learning @cite_30 @cite_52 where it is also known as differentiable programming', and is responsible for enabling efficient inference in many probabilistic programming frameworks @cite_36 @cite_19 @cite_4 @cite_17 @cite_46 . Different automatic differentiation techniques @cite_31 allow different compromises between flexibility, efficiency, and feature-richness @cite_9 @cite_58 @cite_4 . Automatic differentiation is usually implemented through either operator overloading @cite_36 @cite_58 @cite_19 or source code transformation @cite_4 @cite_18 . Our work, too, relies on automatic differentiation, implemented as source code transformation. However, a novelty of our approach is that instead of using explicit calls or directives to denote parts of code which need to be differentiated, we rely on the type system of Go to selectively differentiate the code relevant for inference, thus combining advantages of both operator overloading and source code transformation.
|
{
"cite_N": [
"@cite_30",
"@cite_31",
"@cite_18",
"@cite_4",
"@cite_36",
"@cite_9",
"@cite_52",
"@cite_19",
"@cite_46",
"@cite_58",
"@cite_17"
],
"mid": [
"2962727772",
"1585773866",
"2891673498",
"2899346917",
"",
"",
"2890411254",
"2799261850",
"2962937106",
"2899771611",
"2964321317"
],
"abstract": [
"Derivatives, mostly in the form of gradients and Hessians, are ubiquitous in machine learning. Automatic differentiation (AD), also called algorithmic differentiation or simply \"auto-diff\", is a family of techniques similar to but more general than backpropagation for efficiently and accurately evaluating derivatives of numeric functions expressed as computer programs. AD is a small but established field with applications in areas including computational uid dynamics, atmospheric sciences, and engineering design optimization. Until very recently, the fields of machine learning and AD have largely been unaware of each other and, in some cases, have independently discovered each other's results. Despite its relevance, general-purpose AD has been missing from the machine learning toolbox, a situation slowly changing with its ongoing adoption under the names \"dynamic computational graphs\" and \"differentiable programming\". We survey the intersection of AD and machine learning, cover applications where AD has direct relevance, and address the main implementation techniques. By precisely defining the main differentiation techniques and their interrelationships, we aim to bring clarity to the usage of the terms \"autodiff\", \"automatic differentiation\", and \"symbolic differentiation\" as these are encountered more and more in machine learning settings.",
"Algorithmic, or automatic, differentiation (AD) is a growing area of theoretical research and software development concerned with the accurate and efficient evaluation of derivatives for function evaluations given as computer programs. The resulting derivative values are useful for all scientific computations that are based on linear, quadratic, or higher order approximations to nonlinear scalar or vector functions. AD has been applied in particular to optimization, parameter identification, nonlinear equation solving, the numerical integration of differential equations, and combinations of these. Apart from quantifying sensitivities numerically, AD also yields structural dependence information, such as the sparsity pattern and generic rank of Jacobian matrices. The field opens up an exciting opportunity to develop new algorithms that reflect the true cost of accurate derivatives and to use them for improvements in speed and reliability. This second edition has been updated and expanded to cover recent developments in applications and theory, including an elegant NP completeness argument by Uwe Naumann and a brief introduction to scarcity, a generalization of sparsity. There is also added material on checkpointing and iterative differentiation. To improve readability the more detailed analysis of memory and complexity bounds has been relegated to separate, optional chapters.The book consists of three parts: a stand-alone introduction to the fundamentals of AD and its software; a thorough treatment of methods for sparse problems; and final chapters on program-reversal schedules, higher derivatives, nonsmooth problems and iterative processes. Each of the 15 chapters concludes with examples and exercises. Audience: This volume will be valuable to designers of algorithms and software for nonlinear computational problems. Current numerical software users should gain the insight necessary to choose and deploy existing AD software tools to the best advantage. Contents: Rules; Preface; Prologue; Mathematical Symbols; Chapter 1: Introduction; Chapter 2: A Framework for Evaluating Functions; Chapter 3: Fundamentals of Forward and Reverse; Chapter 4: Memory Issues and Complexity Bounds; Chapter 5: Repeating and Extending Reverse; Chapter 6: Implementation and Software; Chapter 7: Sparse Forward and Reverse; Chapter 8: Exploiting Sparsity by Compression; Chapter 9: Going beyond Forward and Reverse; Chapter 10: Jacobian and Hessian Accumulation; Chapter 11: Observations on Efficiency; Chapter 12: Reversal Schedules and Checkpointing; Chapter 13: Taylor and Tensor Coefficients; Chapter 14: Differentiation without Differentiability; Chapter 15: Implicit and Iterative Differentiation; Epilogue; List of Figures; List of Tables; Assumptions and Definitions; Propositions, Corollaries, and Lemmas; Bibliography; Index",
"The use of derivatives, especially gradients, is pervasive in machine learning, and researchers have access to a wide variety of tools to automatically compute derivatives. However, the need to efficiently calculate first- and higher-order derivatives of increasingly complex models expressed in Python has stressed or exceeded the capabilities of available tools. In this work, we show that techniques from the field of automatic differentiation can give researchers expressive power, performance and strong usability. We implement these ideas in the Tangent software library, an automatic differentiation (AD) framework for Python which uses source-code transformation (SCT) to produce derivatives of user-specified numeric code.",
"Machine learning as a discipline has seen an incredible surge of interest in recent years due in large part to a perfect storm of new theory, superior tooling, renewed interest in its capabilities. We present in this paper a framework named Flux that shows how further refinement of the core ideas of machine learning, built upon the foundation of the Julia programming language, can yield an environment that is simple, easily modifiable, and performant. We detail the fundamental principles of Flux as a framework for differentiable programming, give examples of models that are implemented within Flux to display many of the language and framework-level features that contribute to its ease of use and high productivity, display internal compiler techniques used to enable the acceleration and performance that lies at the heart of Flux, and finally give an overview of the larger ecosystem that Flux fits inside of.",
"",
"",
"We review the current state of automatic differentiation (AD) for array programming in machine learning (ML), including the different approaches such as operator overloading (OO) and source transformation (ST) used for AD, graph-based intermediate representations for programs, and source languages. Based on these insights, we introduce a new graph-based intermediate representation (IR) closely related to A-normal form (ANF) which is specifically aimed at supporting fully-general AD for array programming efficiently. Unlike existing dataflow programming representations in ML frameworks, our intermediate representation (IR) naturally supports function calls, higher-order functions, recursion, etc. making ML models easier to implement. The ability to represent closures allows us to perform AD using ST without a tape, making the resulting derivative (adjoint) program amenable to ahead-of-time optimization using tools from functional language compilers, and enabling higher-order derivatives. Lastly, we introduce a proof of concept compiler toolchain called Myia which uses a subset of Python as a front end.",
"",
"We propose Edward, a Turing-complete probabilistic programming language. Edward defines two compositional representations—random variables and inference. By treating inference as a first class citizen, on a par with modeling, we show that probabilistic programming can be as flexible and computationally efficient as traditional deep learning. For flexibility, Edward makes it easy to fit the same model using a variety of composable inference methods, ranging from point estimation to variational inference to MCMC. In addition, Edward can reuse the modeling representation as part of inference, facilitating the design of rich variational models and generative adversarial networks. For efficiency, Edward is integrated into TensorFlow, providing significant speedups over existing probabilistic systems. For example, we show on a benchmark logistic regression task that Edward is at least 35x faster than Stan and 6x faster than PyMC3. Further, Edward incurs no runtime overhead: it is as fast as handwritten TensorFlow.",
"",
""
]
}
|
1811.09575
|
2901040731
|
In recent years, the sequence-to-sequence learning neural networks with attention mechanism have achieved great progress. However, there are still challenges, especially for Neural Machine Translation (NMT), such as lower translation quality on long sentences. In this paper, we present a hierarchical deep neural network architecture to improve the quality of long sentences translation. The proposed network embeds sequence-to-sequence neural networks into a two-level category hierarchy by following the coarse-to-fine paradigm. Long sentences are input by splitting them into shorter sequences, which can be well processed by the coarse category network as the long distance dependencies for short sentences is able to be handled by network based on sequence-to-sequence neural network. Then they are concatenated and corrected by the fine category network. The experiments shows that our method can achieve superior results with higher BLEU(Bilingual Evaluation Understudy) scores, lower perplexity and better performance in imitating expression style and words usage than the traditional networks.
|
In 1997, A idea of using an “encoder-decoder” structure for machine translation @cite_16 is proposed. A few years later, in 2003, a new proposed neural network-based language model @cite_21 improved the data sparsity of traditional SMT models. Their research lays the foundation for the future application of neural networks in machine translation.
|
{
"cite_N": [
"@cite_16",
"@cite_21"
],
"mid": [
"2172166122",
"2132339004"
],
"abstract": [
"Many researchers have explored the relation between discrete-time recurrent neural networks (DTRNN) and finite-state machines (FSMs) either by showing their computational equivalence or by training them to perform as finite-state recognizers from examples. Most of this work has focused on the simplest class of deterministic state machines, that is deterministic finite automata and Mealy (or Moore) machines. The class of translations these machines can perform is very limited, mainly because these machines output symbols at the same rate as they input symbols, and therefore, the input and the translation have the same length; one may call these translations synchronous. Real-life translations are more complex: word reorderings, deletions, and insertions are common in natural-language translations; or, in speech-to-phoneme conversion, the number of frames corresponding to each phoneme is different and depends on the particular speaker or word. There are, however, simple deterministic, finite-state machines (extensions of Mealy machines) that may perform these classes of \"asynchronous\" or \"time-warped\" translations. A simple DTRNN model with input and output control lines inspired on this class of machines is presented and successfully applied to simple asynchronous translation tasks with interesting results regarding generalization. Training of these nets from input-output pairs is complicated by the fact that the time alignment between the target output sequence and the input sequence is unknown and has to be learned: we propose a new error function to tackle this problem. This approach to the induction of asynchronous translators is discussed in connection with other approaches.",
"A goal of statistical language modeling is to learn the joint probability function of sequences of words in a language. This is intrinsically difficult because of the curse of dimensionality: a word sequence on which the model will be tested is likely to be different from all the word sequences seen during training. Traditional but very successful approaches based on n-grams obtain generalization by concatenating very short overlapping sequences seen in the training set. We propose to fight the curse of dimensionality by learning a distributed representation for words which allows each training sentence to inform the model about an exponential number of semantically neighboring sentences. The model learns simultaneously (1) a distributed representation for each word along with (2) the probability function for word sequences, expressed in terms of these representations. Generalization is obtained because a sequence of words that has never been seen before gets high probability if it is made of words that are similar (in the sense of having a nearby representation) to words forming an already seen sentence. Training such large models (with millions of parameters) within a reasonable time is itself a significant challenge. We report on experiments using neural networks for the probability function, showing on two text corpora that the proposed approach significantly improves on state-of-the-art n-gram models, and that the proposed approach allows to take advantage of longer contexts."
]
}
|
1811.09575
|
2901040731
|
In recent years, the sequence-to-sequence learning neural networks with attention mechanism have achieved great progress. However, there are still challenges, especially for Neural Machine Translation (NMT), such as lower translation quality on long sentences. In this paper, we present a hierarchical deep neural network architecture to improve the quality of long sentences translation. The proposed network embeds sequence-to-sequence neural networks into a two-level category hierarchy by following the coarse-to-fine paradigm. Long sentences are input by splitting them into shorter sequences, which can be well processed by the coarse category network as the long distance dependencies for short sentences is able to be handled by network based on sequence-to-sequence neural network. Then they are concatenated and corrected by the fine category network. The experiments shows that our method can achieve superior results with higher BLEU(Bilingual Evaluation Understudy) scores, lower perplexity and better performance in imitating expression style and words usage than the traditional networks.
|
In 2013, a new end-to-end encoder-decoder architecture for machine translation @cite_8 is introduced. It uses Convolutional Neural Network (CNN) to encode a given piece of source text into a continuous vector, and then use a Recurrent Neural Network (RNN) as a decoder to transform the state vector into a target language, which can be said to be the birth of NMT.
|
{
"cite_N": [
"@cite_8"
],
"mid": [
"1753482797"
],
"abstract": [
"We introduce a class of probabilistic continuous translation models called Recurrent Continuous Translation Models that are purely based on continuous representations for words, phrases and sentences and do not rely on alignments or phrasal translation units. The models have a generation and a conditioning aspect. The generation of the translation is modelled with a target Recurrent Language Model, whereas the conditioning on the source sentence is modelled with a Convolutional Sentence Model. Through various experiments, we show first that our models obtain a perplexity with respect to gold translations that is > 43 lower than that of stateof-the-art alignment-based translation models. Secondly, we show that they are remarkably sensitive to the word order, syntax, and meaning of the source sentence despite lacking alignments. Finally we show that they match a state-of-the-art system when rescoring n-best lists of translations."
]
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.