_id
stringlengths 4
10
| text
stringlengths 0
18.4k
| title
stringlengths 0
8.56k
|
|---|---|---|
d229923250
|
Encoder layer fusion (EncoderFusion) is a technique to fuse all the encoder layers (instead of the uppermost layer) for sequence-to-sequence (Seq2Seq) models, which has proven effective on various NLP tasks. However, it is still not entirely clear why and when EncoderFusion should work. In this paper, our main contribution is to take a step further in understanding EncoderFusion. Many of previous studies believe that the success of EncoderFusion comes from exploiting surface and syntactic information embedded in lower encoder layers. Unlike them, we find that the encoder embedding layer is more important than other intermediate encoder layers. In addition, the uppermost decoder layer consistently pays more attention to the encoder embedding layer across NLP tasks. Based on this observation, we propose a simple fusion method, SurfaceFusion, by fusing only the encoder embedding layer for the softmax layer. Experimental results show that SurfaceFusion outperforms EncoderFusion on several NLP benchmarks, including machine translation, text summarization, and grammatical error correction. It obtains the state-of-the-art performance on WMT16 Romanian-English and WMT14 English-French translation tasks. Extensive analyses reveal that SurfaceFusion learns more expressive bilingual word embeddings by building a closer relationship between relevant source and target embeddings. The source code will be released. * Work was done when Xuebo Liu and Liang Ding were interning at Tencent AI Lab. . Layer-wise cross-view decoding for sequence-to-sequence learning. arXiv, 2020.
|
UNDERSTANDING AND IMPROVING ENCODER LAYER FUSION IN SEQUENCE-TO-SEQUENCE LEARNING
|
d212756
|
Recurrent neural networks (RNNs), such as long short-term memory networks (LSTMs), serve as a fundamental building block for many sequence learning tasks, including machine translation, language modeling, and question answering. In this paper, we consider the specific problem of word-level language modeling and investigate strategies for regularizing and optimizing LSTMbased models. We propose the weight-dropped LSTM which uses DropConnect on hidden-tohidden weights as a form of recurrent regularization. Further, we introduce NT-ASGD, a variant of the averaged stochastic gradient method, wherein the averaging trigger is determined using a non-monotonic condition as opposed to being tuned by the user. Using these and other regularization strategies, we achieve state-of-the-art word level perplexities on two data sets: 57.3 on Penn Treebank and 65.8 on WikiText-2. In exploring the effectiveness of a neural cache in conjunction with our proposed model, we achieve an even lower state-of-the-art perplexity of 52.8 on Penn Treebank and 52.0 on WikiText-2.
|
Regularizing and Optimizing LSTM Language Models
|
d219969405
|
Few-shot classification aims to recognize unseen classes when presented with only a small number of samples. We consider the problem of multi-domain few-shot image classification, where unseen classes and examples come from diverse data sources. This problem has seen growing interest and has inspired the development of benchmarks such as Meta-Dataset. A key challenge in this multi-domain setting is to effectively integrate the feature representations from the diverse set of training domains. Here, we propose a Universal Representation Transformer (URT) layer, that meta-learns to leverage universal features for few-shot classification by dynamically re-weighting and composing the most appropriate domain-specific representations. In experiments, we show that URT sets a new state-of-the-art result on Meta-Dataset. Specifically, it outperforms the best previous model on three data sources or performs the same in others. We analyze variants of URT and present a visualization of the attention score heatmaps that sheds light on how the model performs cross-domain generalization. Our code is available at https://github.com/liulu112601/URT * This work was done while Lu Liu was a research intern with Mila. † Canada CIFAR AI Chair Preprint. Under review.
|
A Universal Representation Transformer Layer for Few-Shot Image Classification
|
d59279266
|
We present graph wavelet neural network (GWNN), a novel graph convolutional neural network (CNN), leveraging graph wavelet transform to address the shortcomings of previous spectral graph CNN methods that depend on graph Fourier transform. Different from graph Fourier transform, graph wavelet transform can be obtained via a fast algorithm without requiring matrix eigendecomposition with high computational cost. Moreover, graph wavelets are sparse and localized in vertex domain, offering high efficiency and good interpretability for graph convolution. The proposed GWNN significantly outperforms previous spectral graph CNNs in the task of graph-based semi-supervised classification on three benchmark datasets: Cora, Citeseer and Pubmed.
|
GRAPH WAVELET NEURAL NETWORK
|
d252815987
|
Building on recent advances in image generation, we present a fully data-driven approach to rendering markup into images. The approach is based on diffusion models, which parameterize the distribution of data using a sequence of denoising operations on top of a Gaussian noise distribution. We view the diffusion denoising process as a sequential decision making process, and show that it exhibits compounding errors similar to exposure bias issues in imitation learning problems. To mitigate these issues, we adapt the scheduled sampling algorithm to diffusion training. We conduct experiments on four markup datasets: mathematical formulas (LaTeX), table layouts (HTML), sheet music (LilyPond), and molecular images (SMILES). These experiments each verify the effectiveness of the diffusion process and the use of scheduled sampling to fix generation issues. These results also show that the markup-to-image task presents a useful controlled compositional setting for diagnosing and analyzing generative image models.
|
MARKUP-TO-IMAGE DIFFUSION MODELS WITH SCHEDULED SAMPLING
|
d209314627
|
Uncertainty estimation and ensembling methods go hand-in-hand. Uncertainty estimation is one of the main benchmarks for assessment of ensembling performance. At the same time, deep learning ensembles have provided state-of-the-art results in uncertainty estimation. In this work, we focus on in-domain uncertainty for image classification. We explore the standards for its quantification and point out pitfalls of existing metrics. Avoiding these pitfalls, we perform a broad study of different ensembling techniques. To provide more insight in this study, we introduce the deep ensemble equivalent score (DEE) and show that many sophisticated ensembling techniques are equivalent to an ensemble of only few independently trained networks in terms of test performance.Published as a conference paper at ICLR 2020 calibrated log-likelihood avoids most of the stated pitfalls and generally is a reasonable metric for in-domain uncertainty estimation task.Published as a conference paper at ICLR 2020 While ensembling techniques tend to have better temperature than single models, the default choice of T = 1 is still suboptimal. Comparing the LL with suboptimal temperatures-that is often the case in practice-can potentially produce an arbitrary ranking of different methods.Comparison of the log-likelihood should only be performed at the optimal temperature.Empirically, we demonstrate that the overall ordering of methods and also the best ensembling method according to the LL can vary depending on temperature T . While this applies to most Timur Garipov, Pavel Izmailov, Dmitrii Podoprikhin, Dmitry P Vetrov, and Andrew G Wilson. Loss surfaces, mode connectivity, and fast ensembling of dnns. son. Averaging weights leads to wider optima and better generalization. arXiv preprint arXiv:1803.05407, 2018.Durk P Kingma, Tim Salimans, and Max Welling. Variational dropout and the local reparameterization trick. . Accuracy-rejection curves (arcs) for comparing classification methods with a reject option. In Machine Learning in Systems Biology, pp. 65-81, 2009.Mahdi Pakdaman Naeini, Gregory Cooper, and Milos Hauskrecht. Obtaining well calibrated probabilities using bayesian binning. In
|
PITFALLS OF IN-DOMAIN UNCERTAINTY ESTIMATION AND ENSEMBLING IN DEEP LEARNING
|
d203737303
|
Deep neural networks (DNNs) have been shown to over-fit a dataset when being trained with noisy labels for a long enough time. To overcome this problem, we present a simple and effective method self-ensemble label filtering (SELF) to progressively filter out the wrong labels during training. Our method improves the task performance by gradually allowing supervision only from the potentially non-noisy (clean) labels and stops learning on the filtered noisy labels. For the filtering, we form running averages of predictions over the entire training dataset using the network output at different training epochs. We show that these ensemble estimates yield more accurate identification of inconsistent predictions throughout training than the single estimates of the network at the most recent training epoch. While filtered samples are removed entirely from the supervised training loss, we dynamically leverage them via semi-supervised learning in the unsupervised loss. We demonstrate the positive effect of such an approach on various image classification tasks under both symmetric and asymmetric label noise and at different noise ratios. It substantially outperforms all previous works on noise-aware learning across different datasets and can be applied to a broad set of network architectures.
|
SELF: LEARNING TO FILTER NOISY LABELS WITH SELF-ENSEMBLING
|
d222208810
|
Given a simple request (e.g., Put a washed apple in the kitchen fridge), humans can reason in purely abstract terms by imagining action sequences and scoring their likelihood of success, prototypicality, and efficiency, all without moving a muscle.Once we see the kitchen in question, we can update our abstract plans to fit the scene. Embodied agents require the same abilities, but existing work does not yet provide the infrastructure necessary for both reasoning abstractly and executing concretely. We address this limitation by introducing ALFWorld, a simulator that enables agents to learn abstract, text-based policies in TextWorld(Côté et al., 2018)and then execute goals from the ALFRED benchmark (Shridhar et al., 2020) in a rich visual environment. ALFWorld enables the creation of a new BUTLER agent whose abstract knowledge, learned in TextWorld, corresponds directly to concrete, visually grounded actions. In turn, as we demonstrate empirically, this fosters better agent generalization than training only in the visually grounded environment. BUTLER's simple, modular design factors the problem to allow researchers to focus on models for improving every piece of the pipeline (language understanding, planning, navigation, visual scene understanding, and so forth). Data, code, and videos are available at: alfworld.github.io arXiv:2010.03768v1 [cs.CL]
|
ALFWORLD: ALIGNING TEXT AND EMBODIED ENVIRONMENTS FOR INTERACTIVE LEARNING
|
d255570226
|
The combinatorial pure exploration of causal bandits is the following online learning task: given a causal graph with unknown causal inference distributions, in each round we choose a subset of variables to intervene or do no intervention, and observe the random outcomes of all random variables, with the goal that using as few rounds as possible, we can output an intervention that gives the best (or almost best) expected outcome on the reward variable Y with probability at least 1 − δ, where δ is a given confidence level. We provide the first gap-dependent and fully adaptive pure exploration algorithms on two types of causal models -the binary generalized linear model (BGLM) and general graphs. For BGLM, our algorithm is the first to be designed specifically for this setting and achieves polynomial sample complexity, while all existing algorithms for general graphs have either sample complexity exponential to the graph size or some unreasonable assumptions. For general graphs, our algorithm provides a significant improvement on sample complexity, and it nearly matches the lower bound we prove. Our algorithms achieve such improvement by a novel integration of prior causal bandit algorithms and prior adaptive pure exploration algorithms, the former of which utilize the rich observational feedback in causal bandits but are not adaptive to reward gaps, while the latter of which have the issue in reverse.
|
Combinatorial Pure Exploration of Causal Bandits
|
d104292008
|
Due to the lack of enough training data and high computational cost to train a deep neural network from scratch, transfer learning has been extensively used in many deep-neural-network-based applications, such as face recognition, image classification, speech recognition, etc. A commonly-used transfer learning approach involves taking a part of a pre-trained model, adding a few layers at the end, and re-training the new layers with a small dataset. This approach, while efficient and widely used, imposes a security vulnerability because the pre-trained models used in transfer learning are usually available publicly to everyone, including potential attackers. In this paper, we show that without any additional knowledge other than the pre-trained model, an attacker can launch an effective and efficient brute force attack that can craft instances of input to trigger each target class with high confidence. Note that we assume that the attacker does not have access to any targetspecific information, including samples from target classes, re-trained model, and probabilities assigned by Softmax to each class, and thus called target-agnostic attack. These assumptions render all previous attacks impractical, to the best of our knowledge. To evaluate the proposed attack, we perform a set of experiments on face recognition and speech recognition tasks and show the effectiveness of the attack. Our work sheds light on a fundamental security challenge of transfer learning in deep neural networks.
|
A Target-Agnostic Attack on Deep Models: Exploiting Security Vulnerabilities of Transfer Learning
|
d239024909
|
Visual search, recommendation, and contrastive similarity learning power technologies that impact billions of users worldwide. Modern model architectures can be complex and difficult to interpret, and there are several competing techniques one can use to explain a search engine's behavior. We show that the theory of fair credit assignment provides a unique axiomatic solution that generalizes several existing recommendation-and metric-explainability techniques in the literature. Using this formalism, we show when existing approaches violate "fairness" and derive methods that sidestep these shortcomings and naturally handle counterfactual information. More specifically, we show existing approaches implicitly approximate second-order Shapley-Taylor indices and extend CAM, GradCAM, LIME, SHAP, SBSM, and other methods to search engines. These extensions can extract pairwise correspondences between images from trained opaque-box models. We also introduce a fast kernel-based method for estimating Shapley-Taylor indices that require orders of magnitude fewer function evaluations to converge. Finally, we show that these game-theoretic measures yield more consistent explanations for image similarity architectures.
|
AXIOMATIC EXPLANATIONS FOR VISUAL SEARCH, RETRIEVAL, AND SIMILARITY LEARNING
|
d260126025
|
Robust reinforcement learning (RL) seeks to train policies that can perform well under environment perturbations or adversarial attacks. Existing approaches typically assume that the space of possible perturbations remains the same across timesteps. However, in many settings, the space of possible perturbations at a given timestep depends on past perturbations. We formally introduce temporally-coupled perturbations, presenting a novel challenge for existing robust RL methods. To tackle this challenge, we propose GRAD, a novel game-theoretic approach that treats the temporally-coupled robust RL problem as a partially-observable twoplayer zero-sum game. By finding an approximate equilibrium in this game, GRAD ensures the agent's robustness against temporally-coupled perturbations. Empirical experiments on a variety of continuous control tasks demonstrate that our proposed approach exhibits significant robustness advantages compared to baselines against both standard and temporally-coupled attacks, in both state and action spaces.
|
Game-Theoretic Robust Reinforcement Learning Handles Temporally-Coupled Perturbations
|
d3524564
|
We present an end-to-end trained memory system that quickly adapts to new data and generates samples like them. Inspired by Kanerva's sparse distributed memory, it has a robust distributed reading and writing mechanism. The memory is analytically tractable, which enables optimal on-line compression via a Bayesian update-rule. We formulate it as a hierarchical conditional generative model, where memory provides a rich data-dependent prior distribution. Consequently, the top-down memory and bottom-up perception are combined to produce the code representing an observation. Empirically, we demonstrate that the adaptive memory significantly improves generative models trained on both the Omniglot and CIFAR datasets. Compared with the Differentiable Neural Computer (DNC) and its variants, our memory model has greater capacity and is significantly easier to train.
|
THE KANERVA MACHINE: A GENERATIVE DISTRIBUTED MEMORY
|
d226237047
|
State-of-the-art natural language understanding classification models follow twostages: pre-training a large language model on an auxiliary task, and then finetuning the model on a task-specific labeled dataset using cross-entropy loss. Crossentropy loss has several shortcomings that can lead to sub-optimal generalization and instability. Driven by the intuition that good generalization requires capturing the similarity between examples in one class and contrasting them with examples in other classes, we propose a supervised contrastive learning (SCL) objective for the fine-tuning stage. Combined with cross-entropy, the SCL loss we propose obtains improvements over a strong RoBERTa-Large baseline on multiple datasets of the GLUE benchmark in both the high-data and low-data regimes, and it does not require any specialized architecture, data augmentation of any kind, memory banks, or additional unsupervised data. We also demonstrate that the new objective leads to models that are more robust to different levels of noise in the training data, and can generalize better to related tasks with limited labeled task data.
|
SUPERVISED CONTRASTIVE LEARNING FOR PRE-TRAINED LANGUAGE MODEL FINE-TUNING
|
d52900371
|
The notion of the stationary equilibrium ensemble has played a central role in statistical mechanics. In machine learning as well, training serves as generalized equilibration that drives the probability distribution of model parameters toward stationarity. Here, we derive stationary fluctuation-dissipation relations that link measurable quantities and hyperparameters in the stochastic gradient descent algorithm. These relations hold exactly for any stationary state and can in particular be used to adaptively set training schedule. We can further use the relations to efficiently extract information pertaining to a loss-function landscape such as the magnitudes of its Hessian and anharmonicity. Our claims are empirically verified.
|
FLUCTUATION-DISSIPATION RELATIONS FOR STOCHASTIC GRADIENT DESCENT
|
d226254365
|
Effective training of deep neural networks can be challenging, and there remain many open questions on how to best learn these models. Recently developed methods to improve neural network training examine teaching: providing learned information during the training process to improve downstream model performance. In this paper, we take steps towards extending the scope of teaching. We propose a flexible teaching framework using commentaries, meta-learned information helpful for training on a particular task or dataset. We present an efficient and scalable gradient-based method to learn commentaries, leveraging recent work on implicit differentiation. We explore diverse applications of commentaries, from learning weights for individual training examples, to parameterizing label-dependent data augmentation policies, to representing attention masks that highlight salient image regions. In these settings, we find that commentaries can improve training speed and/or performance and also provide fundamental insights about the dataset and training process. * Correspondence to: aniruddhraghu@gmail.com. Work done while interning at Google. arXiv:2011.03037v1 [cs.LG] 5 Nov 2020 4. We investigate label-dependent commentaries to define a data augmentation policy, and obtain insights into the design of effective augmentation strategies and improved performance on benchmark tasks as compared to baselines. 5. We parameterize commentaries as attention masks to find important regions of images. Through qualitative and quantitative evaluation, we show these masks identify salient image regions and can be used to improve the robustness of neural networks to spurious background correlations.Teaching with CommentariesDefinition: We define a commentary to be learned information helpful for (i) training a model on a task or (ii) providing insights on the learning process. Formally, let t(x, y, i; φ) denote a commentary that is a function of a data point x, prediction target y, and iteration of training i, with parameters φ. The commentary may be represented in a tabular fashion for every combination of input arguments, or using a neural network that takes these arguments as inputs.The commentary is used in the training of a student network with parameters θ, denoted n(x; θ).
|
Teaching with Commentaries
|
d245836975
|
We present LSeg, a novel model for language-driven semantic image segmentation. LSeg uses a text encoder to compute embeddings of descriptive input labels (e.g., "grass" or "building") together with a transformer-based image encoder that computes dense per-pixel embeddings of the input image. The image encoder is trained with a contrastive objective to align pixel embeddings to the text embedding of the corresponding semantic class. The text embeddings provide a flexible label representation in which semantically similar labels map to similar regions in the embedding space (e.g., "cat" and "furry"). This allows LSeg to generalize to previously unseen categories at test time, without retraining or even requiring a single additional training sample. We demonstrate that our approach achieves highly competitive zero-shot performance compared to existing zero-and few-shot semantic segmentation methods, and even matches the accuracy of traditional segmentation algorithms when a fixed label set is provided. Code and demo are available at https://github.com/isl-org/lang-seg.
|
LANGUAGE-DRIVEN SEMANTIC SEGMENTATION
|
d263829725
|
Topology reasoning aims to comprehensively understand road scenes and present drivable routes in autonomous driving.It requires detecting road centerlines (lane) and traffic elements, further reasoning their topology relationship, i.e., lane-lane topology, and lane-traffic topology.In this work, we first present that the topology score relies heavily on detection performance on lane and traffic elements.Therefore, we introduce a powerful 3D lane detector and an improved 2D traffic element detector to extend the upper limit of topology performance.Further, we propose TopoMLP, a simple yet high-performance pipeline for driving topology reasoning.Based on the impressive detection performance, we develop two simple MLP-based heads for topology generation.TopoMLP achieves state-of-the-art performance on OpenLane-V2 benchmark, i.e., 41.2% OLS with ResNet-50 backbone.It is also the 1st solution for 1st OpenLane Topology in Autonomous Driving Challenge.We hope such simple and strong pipeline can provide some new insights to the community.Code is at https://github.com/wudongming97/TopoMLP.
|
TOPOMLP: A SIMPLE YET STRONG PIPELINE FOR DRIVING TOPOLOGY REASONING
|
d52890982
|
While Generative Adversarial Networks (GANs) have seen wide success at the problem of synthesizing realistic images, they have seen little application to audio generation. Unlike for images, a barrier to success is that the best discriminative representations for audio tend to be non-invertible, and thus cannot be used to synthesize listenable outputs. In this paper we introduce WaveGAN, a first attempt at applying GANs to unsupervised synthesis of raw-waveform audio. Our experiments demonstrate that WaveGAN can produce intelligible words from a small vocabulary of speech, and can also synthesize audio from other domains such as drums, bird vocalizations, and piano. Qualitatively, we find that human judges prefer the sound quality of generated examples from WaveGAN over those from a method which naïvely apply GANs on image-like audio feature representations.
|
ADVERSARIAL AUDIO SYNTHESIS
|
d85459724
|
As people learn to navigate the world, autonomic nervous system (e.g., "fight or flight") responses provide intrinsic feedback about the potential consequence of action choices (e.g., becoming nervous when close to a cliff edge or driving fast around a bend.) Physiological changes are correlated with these biological preparations to protect one-self from danger. We present a novel approach to reinforcement learning that leverages a task-independent intrinsic reward function trained on peripheral pulse measurements that are correlated with human autonomic nervous system responses. Our hypothesis is that such reward functions can circumvent the challenges associated with sparse and skewed rewards in reinforcement learning settings and can help improve sample efficiency. We test this in a simulated driving environment and show that it can increase the speed of learning and reduce the number of collisions during the learning stage.
|
VISCERAL MACHINES: RISK-AVERSION IN REINFORCEMENT LEARNING WITH INTRINSIC PHYSIOLOGICAL REWARDS
|
d256416405
|
In recent years numerous methods have been developed to formally verify the robustness of deep neural networks (DNNs). Though the proposed techniques are effective in providing mathematical guarantees about the DNNs behavior, it is not clear whether the proofs generated by these methods are human-interpretable. In this paper, we bridge this gap by developing new concepts, algorithms, and representations to generate human understandable interpretations of the proofs. Leveraging the proposed method, we show that the robustness proofs of standard DNNs rely on spurious input features, while the proofs of DNNs trained to be provably robust filter out even the semantically meaningful features. The proofs for the DNNs combining adversarial and provably robust training are the most effective at selectively filtering out spurious features as well as relying on human-understandable input features.
|
Interpreting Robustness Proofs of Deep Neural Networks
|
d249097375
|
Designing robust loss functions is popular in learning with noisy labels while existing designs did not explicitly consider the overfitting property of deep neural networks (DNNs). As a result, applying these losses may still suffer from overfitting/memorizing noisy labels as training proceeds. In this paper, we first theoretically analyze the memorization effect and show that a lower-capacity model may perform better on noisy datasets. However, it is non-trivial to design a neural network with the best capacity given an arbitrary task. To circumvent this dilemma, instead of changing the model architecture, we decouple DNNs into an encoder followed by a linear classifier and propose to restrict the function space of a DNN by a representation regularizer. Particularly, we require the distance between two self-supervised features to be positively related to the distance between the corresponding two supervised model outputs. Our proposed framework is easily extendable and can incorporate many other robust loss functions to further improve performance. Extensive experiments and theoretical analyses support our claims. Code is available at github.com/UCSC-REAL/SelfSup_NoisyLabel.
|
Mitigating Memorization of Noisy Labels via Regularization between Representations
|
d189898449
|
Network pruning is a promising avenue for compressing deep neural networks. A typical approach to pruning starts by training a model and then removing redundant parameters while minimizing the impact on what is learned. Alternatively, a recent approach shows that pruning can be done at initialization prior to training, based on a saliency criterion called connection sensitivity. However, it remains unclear exactly why pruning an untrained, randomly initialized neural network is effective. In this work, by noting connection sensitivity as a form of gradient, we formally characterize initialization conditions to ensure reliable connection sensitivity measurements, which in turn yields effective pruning results. Moreover, we analyze the signal propagation properties of the resulting pruned networks and introduce a simple, data-free method to improve their trainability. Our modifications to the existing pruning at initialization method lead to improved results on all tested network models for image classification tasks. Furthermore, we empirically study the effect of supervision for pruning and demonstrate that our signal propagation perspective, combined with unsupervised pruning, can be useful in various scenarios where pruning is applied to non-standard arbitrarily-designed architectures.
|
A SIGNAL PROPAGATION PERSPECTIVE FOR PRUNING NEURAL NETWORKS AT INITIALIZATION
|
d246442105
|
Continuous-depth neural networks, such as Neural ODEs, have refashioned the understanding of residual neural networks in terms of non-linear vector-valued optimal control problems. The common solution is to use the adjoint sensitivity method to replicate a forward-backward pass optimisation problem. We propose a new approach which explicates the network's 'depth' as a fundamental variable, thus reducing the problem to a system of forward-facing initial value problems. This new method is based on the principle of 'Invariant Imbedding' for which we prove a general solution, applicable to all non-linear, vector-valued optimal control problems with both running and terminal loss. Our new architectures provide a tangible tool for inspecting the theoretical-and to a great extent unexplainedproperties of network depth. They also constitute a resource of discrete implementations of Neural ODEs comparable to classes of imbedded residual neural networks. Through a series of experiments, we show the competitive performance of the proposed architectures for supervised learning and time series prediction. Accompanying code is made available at github.com/andrw3000/inimnet. * Equal Contribution.Published as a conference paper at ICLR 2022 perceptrons. But following the advent of residual neural networks(He et al., 2015)which use 'Euler-step' internal updates between layers, DNN evolution is seen to emulate a continuous dynamical system(Lu et al., 2018;Ruthotto & Haber, 2020). Thus was formed the notion of a 'Neural ODE'(Chen et al., 2018)in which the hidden network state vector z(t) ∈ R N , instead of being defined at fixed layers t ∈ N, is allowed to vary continuously over an interval [p, q] ⊂ R for a system of dimension N ≥ 1. Its evolution is governed by an Ordinary Differential Equation (ODE)where the training function f is controlled by a parameter vector θ(t) ∈ R M for t ∈ [p, q] of length M ≥ 1. Network outputs are retrieved at z(q) = y after fixing the endpoint t = q. As such, the enigmatic 'depth' of a Neural ODE is controlled by varying t = p, at which point we insert the initial condition z(p) = x. This dynamical description has given the theory of DNNs a new home: the mathematical framework of optimal control (Massaroli et al., 2020;Bensoussan et al., 2020). In this framework, whether discrete or continuous, solution networks are sought after that: (A) satisfying their update law (1) over a fixed 'depth' interval [p, q]; (B) minimise a loss function subject to terminal success and internal regulation (see §2.1). As initiated by Euler and Lagrange (Euler, 1766), this mathematical framework determines networks given by (A) satisfying condition (B) using, amongst other techniques, the adjoint method: here a new state, which we call λ(t) or the 'Lagrange multiplier', is introduced containing the system losses with respect to both t and the parameters θ(t); see §2.3.The connection between the traditional method of 'backpropagation-by-chain rule' and the rigorous 'adjoint method' is quite brilliantly explicated in proof byChen et al. (2018), in that they directly deduce the adjoint method using infinitesimal backpropagation-far from the train of thought of Lagrange's multiplier method (Liberzon, 2012, Ch. 2); see also §D and §E.2 in the appendix for extended discussion and derivation.Even within the above theory, the initial condition, z(p) = x, and location, t = p, remain implicit constraints; a clear understanding of network depth remains illusive.Our new class of InImNet architectures may be obtained by imbedding networks of varying depth p whilst keeping the inputs, x, invariant. Explicating these two variables throughout the network, writing z(t) = z(t; p, x), has exciting conceptual consequences:
|
IMBEDDING DEEP NEURAL NETWORKS
|
d246607791
|
In this paper, we study the representation of neural networks from the view of kernels. We first define the Neural Fisher Kernel (NFK), which is the Fisher Kernel (Jaakkola and Haussler, 1998) applied to neural networks. We show that NFK can be computed for both supervised and unsupervised learning models, which can serve as a unified tool for representation extraction. Furthermore, we show that practical NFKs exhibit low-rank structures. We then propose an efficient algorithm that computes a low rank approximation of NFK, which scales to large datasets and networks. We show that the low-rank approximation of NFKs derived from unsupervised generative models and supervised learning models gives rise to high-quality compact representations of data, achieving competitive results on a variety of machine learning tasks.Published as a conference paper at ICLR 2022 linear feature space of the kernel. Kernel machines provide drastically different representations than layer activations, where the knowledge of a neural network is instantiated by the induced kernel function over data points.In this work, we propose to make use of the linear feature space of the kernel, associated with the pre-trained neural network model, as the data representation of interest. To this end, we made novel contributions on both theoretical and empirical side, as summarized below.• We propose Neural Fisher Kernel (NFK) as a unified and principled kernel formulation for neural networks models in both supervised learning and unsupervised learning settings.• We introduce a highly efficient and scalable algorithm for low-rank kernel approximation of NFK, which allows us to obtain a compact yet informative feature embedding as the data representation.• We validate the effectiveness of proposed approach from NFK in unsupervised learning, semisupervised learning and supervised learning settings, showing that our method enjoys superior sample efficiency and representation quality.PRELIMINARY AND RELATED WORKSIn this section, we present technical background and formalize the motivation. We start by introducing the notion of data representation from the perspective of kernel methods, then introduce the connections between neural network models and kernel methods.Notations. Throughout this paper, we consider dataset with N
|
LEARNING REPRESENTATION FROM NEURAL FISHER KERNEL WITH LOW-RANK APPROXIMATION
|
d257079072
|
Discovering interpretable patterns for classification of sequential data is of key importance for a variety of fields, ranging from genomics to fraud detection or more generally interpretable decision-making. In this paper, we propose a novel differentiable fully interpretable method to discover both local and global patterns (i.e. catching a relative or absolute temporal dependency) for rule-based binary classification. It consists of a convolutional binary neural network with an interpretable neural filter and a training strategy based on dynamically-enforced sparsity. We demonstrate the validity and usefulness of the approach on synthetic datasets and on an open-source peptides dataset. Key to this end-to-end differentiable method is that the expressive patterns used in the rules are learned alongside the rules themselves.
|
NEURAL-BASED CLASSIFICATION RULE LEARNING FOR SEQUENTIAL DATA
|
d262083735
|
Calibration measures and reliability diagrams are two fundamental tools for measuring and interpreting the calibration of probabilistic predictors.Calibration measures quantify the degree of miscalibration, and reliability diagrams visualize the structure of this miscalibration.However, the most common constructions of reliability diagrams and calibration measures -binning and ECE -both suffer from well-known flaws (e.g.discontinuity).We show that a simple modification fixes both constructions: first smooth the observations using an RBF kernel, then compute the Expected Calibration Error (ECE) of this smoothed function.We prove that with a careful choice of bandwidth, this method yields a calibration measure that is well-behaved in the sense of Błasiok, Gopalan, Hu, and Nakkiran (2023) -a consistent calibration measure.We call this measure the SmoothECE.Moreover, the reliability diagram obtained from this smoothed function visually encodes the SmoothECE, just as binned reliability diagrams encode the BinnedECE.We also provide a Python package with simple, hyperparameter-free methods for measuring and plotting calibration: `pip install relplot`.Code at: https://github.com/apple/ml-calibration.
|
Smooth ECE: Principled Reliability Diagrams via Kernel Smoothing
|
d253107476
|
Despite the empirical advances of deep learning across a variety of learning tasks, our theoretical understanding of its success is still very restricted. One of the key challenges is the overparametrized nature of modern models, enabling complete overfitting of the data even if the labels are randomized, i.e. networks can completely memorize all given patterns. While such a memorization capacity seems worrisome, in this work we show that under training protocols that include data augmentation, neural networks learn to memorize entirely random labels in a benign way, i.e. they learn embeddings that lead to highly non-trivial performance under nearest neighbour probing. We demonstrate that deep models have the surprising ability to separate noise from signal by distributing the task of memorization and feature learning to different layers. As a result, only the very last layers are used for memorization, while preceding layers encode performant features which remain largely unaffected by the label noise. We explore the intricate role of the augmentations used for training and identify a memorization-generalization trade-off in terms of their diversity, marking a clear distinction to all previous works. Finally, we give a first explanation for the emergence of benign memorization by showing that malign memorization under data augmentation is infeasible due to the insufficient capacity of the model for the increased sample size. As a consequence, the network is forced to leverage the correlated nature of the augmentations and as a result learns meaningful features. To complete the picture, a better theory of feature learning in deep neural networks is required to fully understand the origins of this phenomenon. * Equal contribution.
|
THE CURIOUS CASE OF BENIGN MEMORIZATION
|
d220041969
|
We propose a simple, practical, and intuitive approach for domain adaptation in reinforcement learning. Our approach stems from the idea that the agent's experience in the source domain should look similar to its experience in the target domain. Building off of a probabilistic view of RL, we formally show that we can achieve this goal by compensating for the difference in dynamics by modifying the reward function. This modified reward function is simple to estimate by learning auxiliary classifiers that distinguish source-domain transitions from target-domain transitions. Intuitively, the modified reward function penalizes the agent for visiting states and taking actions in the source domain which are not possible in the target domain. Said another way, the agent is penalized for transitions that would indicate that the agent is interacting with the source domain, rather than the target domain. Our approach is applicable to domains with continuous states and actions and does not require learning an explicit model of the dynamics. On discrete and continuous control tasks, we illustrate the mechanics of our approach and demonstrate its scalability to high-dimensional tasks. * Equal contribution.Preprint. Under review.
|
Off-Dynamics Reinforcement Learning: Training for Transfer with Domain Classifiers
|
d2808403
|
We consider the use of Deep Learning methods for modeling complex phenomena like those occurring in natural physical processes. With the large amount of data gathered on these phenomena the data intensive paradigm could begin to challenge more traditional approaches elaborated over the years in fields like maths or physics. However, despite considerable successes in a variety of application domains, the machine learning field is not yet ready to handle the level of complexity required by such problems. Using an example application, namely Sea Surface Temperature Prediction, we show how general background knowledge gained from physics could be used as a guideline for designing efficient Deep Learning models. In order to motivate the approach and to assess its generality we demonstrate a formal link between the solution of a class of differential equations underlying a large family of physical phenomena and the proposed model. Experiments and comparison with series of baselines including a state of the art numerical approach is then provided. * equal contribution .
|
Deep Learning for Physical Processes: Incorporating Prior Scientific Knowledge
|
d220961494
|
A recent line of work addresses the problem of predicting high-dimensional spatiotemporal phenomena by leveraging specific tools from the differential equations theory. Following this direction, we propose in this article a novel and general paradigm for this task based on a resolution method for partial differential equations: the separation of variables. This inspiration allows to introduce a dynamical interpretation of spatiotemporal disentanglement. It induces a simple and principled model based on learning disentangled spatial and temporal representations of a phenomenon to accurately predict future observations. We experimentally demonstrate the performance and broad applicability of our method against prior state-of-the-art models on physical and synthetic video datasets. * Equal contribution.Preprint. Under review.
|
PDE-Driven Spatiotemporal Disentanglement
|
d29154793
|
We present a method for translating music across musical instruments, genres, and styles. This method is based on a multi-domain wavenet autoencoder, with a shared encoder and a disentangled latent space that is trained end-to-end on waveforms. Employing a diverse training dataset and large net capacity, the domain-independent encoder allows us to translate even from musical domains that were not seen during training. The method is unsupervised and does not rely on supervision in the form of matched samples between domains or musical transcriptions. We evaluate our method on NSynth, as well as on a dataset collected from professional musicians, and achieve convincing translations, even when translating from whistling, potentially enabling the creation of instrumental music by untrained humans.
|
A Universal Music Translation Network
|
d247244739
|
Machine learning classifiers with high test accuracy often perform poorly under adversarial attacks. It is commonly believed that adversarial training alleviates this issue. In this paper, we demonstrate that, surprisingly, the opposite may be true -Even though adversarial training helps when enough data is available, it may hurt robust generalization in the small sample size regime. We first prove this phenomenon for a high-dimensional linear classification setting with noiseless observations. Our proof provides explanatory insights that may also transfer to feature learning models. Further, we observe in experiments on standard image datasets that the same behavior occurs for perceptible attacks that effectively reduce class information such as mask attacks and object corruptions.
|
Why adversarial training can hurt robust accuracy
|
d8153918
|
We study the problem of semantic code repair, which can be broadly defined as automatically fixing non-syntactic bugs in source code. The majority of past work in semantic code repair assumed access to unit tests against which candidate repairs could be validated. In contrast, the goal here is to develop a strong statistical model to accurately predict both bug locations and exact fixes without access to information about the intended correct behavior of the program. Achieving such a goal requires a robust contextual repair model, which we train on a large corpus of real-world source code that has been augmented with synthetically injected bugs. Our framework adopts a two-stage approach where first a large set of repair candidates are generated by rule-based processors, and then these candidates are scored by a statistical model using a novel neural network architecture which we refer to as Share, Specialize, and Compete. Specifically, the architecture (1) generates a shared encoding of the source code using an RNN over the abstract syntax tree, (2) scores each candidate repair using specialized network modules, and (3) then normalizes these scores together so they can compete against one another in comparable probability space. We evaluate our model on a real-world test set gathered from GitHub containing four common categories of bugs. Our model is able to predict the exact correct repair 41% of the time with a single guess, compared to 13% accuracy for an attentional sequence-to-sequence model.
|
Semantic Code Repair using Neuro-Symbolic Transformation Networks
|
d238419270
|
Reinforcement learning encounters many challenges when applied directly in the real world. Sim-to-real transfer is widely used to transfer the knowledge learned from simulation to the real world. Domain randomization-one of the most popular algorithms for sim-to-real transfer-has been demonstrated to be effective in various tasks in robotics and autonomous driving. Despite its empirical successes, theoretical understanding on why this simple algorithm works is limited. In this paper, we propose a theoretical framework for sim-to-real transfers, in which the simulator is modeled as a set of MDPs with tunable parameters (corresponding to unknown physical parameters such as friction). We provide sharp bounds on the sim-to-real gap-the difference between the value of policy returned by domain randomization and the value of an optimal policy for the real world. We prove that sim-to-real transfer can succeed under mild conditions without any real-world training samples. Our theory also highlights the importance of using memory (i.e., history-dependent policies) in domain randomization. Our proof is based on novel techniques that reduce the problem of bounding the sim-to-real gap to the problem of designing efficient learning algorithms for infinite-horizon MDPs, which we believe are of independent interest. * These two authors contributed equally. Shipra Agrawal and Randy Jia. Posterior sampling for reinforcement learning: worst-case regret bounds. arXiv preprint arXiv:1705.07041, 2017. , et al. Using simulation and domain adaptation to improve efficiency of deep robotic grasping. In 2018 IEEE international conference on robotics and automation (ICRA), pp. 4243-4250. IEEE, 2018. Peter Buchholz and Dimitri Scheftelowitsch. Computation of weighted sums of rewards for concurrent MDPs. . Transfer from simulation to real world through learning deep inverse dynamics model. arXiv preprint arXiv:1610.03518, 2016.Mark Cutler and Jonathan P How. Efficient reinforcement learning for robots using informative simulated priors. . Sim2real transfer learning for 3d human pose estimation: motion to the rescue.
|
UNDERSTANDING DOMAIN RANDOMIZATION FOR SIM-TO-REAL TRANSFER
|
d222090711
|
We present a novel view on principal component analysis (PCA) as a competitive game in which each approximate eigenvector is controlled by a player whose goal is to maximize their own utility function. We analyze the properties of this PCA game and the behavior of its gradient based updates. The resulting algorithm-which combines elements from Oja's rule with a generalized Gram-Schmidt orthogonalization-is naturally decentralized and hence parallelizable through message passing. We demonstrate the scalability of the algorithm with experiments on large image datasets and neural network activations. We discuss how this new view of PCA as a differentiable game can lead to further algorithmic developments and insights.Preprint. Under review.
|
EigenGame: PCA as a Nash Equilibrium
|
d258686472
|
Recent studies have shown that episodic reinforcement learning (RL) is no harder than bandits when the total reward is bounded by 1, and proved regret bounds that have a polylogarithmic dependence on the planning horizon H. However, it remains an open question that if such results can be carried over to adversarial RL, where the reward is adversarially chosen at each episode. In this paper, we answer this question affirmatively by proposing the first horizon-free policy search algorithm. To tackle the challenges caused by exploration and adversarially chosen reward, our algorithm employs (1) a variance-uncertainty-aware weighted least square estimator for the transition kernel; and (2) an occupancy measure-based technique for the online search of a stochastic policy. We show that our algorithm achieves an O (d + log(|S| 2 |A|)) √ K regret with full-information feedback 2 , where d is the dimension of a known feature mapping linearly parametrizing the unknown transition kernel of the MDP, K is the number of episodes, |S| and |A| are the cardinalities of the state and action spaces. We also provide hardness results and regret lower bounds to justify the near optimality of our algorithm and the unavoidability of log |S| and log |A| in the regret bound. 2 Here O(·) hides logarithmic factors of H, K and 1/δ. 1 arXiv:2305.08359v1 [cs.LG] 15 May 2023 with no polynomial dependence on H for both tabular MDPs (Zhang et al., 2021a) and RL with linear function approximation(Zhang et al., 2021b;Kim et al., 2022;Zhou and Gu, 2022). This suggests that episodic RL is no more difficult than contextual bandits (CB), which is equivalent to episodic RL with H = 1 and no state transition. Nevertheless, these horizon-free algorithms are only applicable to learning MDPs where the reward function is either fixed or stochastic, yet in many real-world scenarios, we have cope with the adversarially changing reward(Even-Dar et al., 2009;Yu et al., 2009;Zimin and Neu, 2013). However, little is known about horizon-free RL in adversarial MDPs. Thus, the following question remains open.Can we design near-optimal horizon-free RL algorithms under adversarial reward and unknown transition with function approximation employed?In this paper, we affirmatively answer the question in the setting of linear mixture MDPs with adversarial reward under full-information feedback(Cai et al., 2020;He et al., 2022b). We propose a new algorithm termed Horizon-Free Occupancy-Measure Guided Optimistic Policy Search (HF-O 2 PS). Following Cai et al. (2020); He et al. (2022b), we use online mirror descent (OMD) to update the policies and value-targeted regression (VTR) (Jia et al., 2020; Ayoub et al., 2020) to learn the transition. Nevertheless, we show that the value-function-based mirror descent inevitably introduces the polynomial dependency on the planning horizon H in the regret upper bound. To address this issue, inspired by Rosenberg and Mansour (2019a); Jin et al. (2020a) and Kalagarla et al. (2020), we use occupancy measure as a proxy of the policy and conduct OMD on the occupancy measures to update. Like Jin et al. (2020a), we maintain a confidence set of the transition kernel and utilize constrained OMD to handle the unknown transition. To achieve a horizon-free regret bound, we also extend the high-order moment estimator inZhou and Gu (2022)to stochastic policies and obtain a tighter Bernstein-type confidence set. The regret of our algorithm can be upper bounded by O(d √ K + d 2 ) in the first K episodes with high probability, where d is the dimension of the feature mapping. To the best of our knowledge, our algorithm is the first algorithm to achieve horizon-free regret in learning adversarial linear mixture MDPs. Our three main contributions are summarized as follows. • We propose a new algorithm, HF-O 2 PS, for linear mixture MDPs with adversarial reward uniformly bounded by 1/H. Compared to the previous works (e.g., Cai et al. (2020); He et al. (2022b)), HF-O 2 PS use occupancy-measure-based OMD rather than direct policy optimization. HF-O 2 PS also use the high-order moment estimator to further facilitate the learning of the transition kernel.• Our analysis shows that HF-O 2 PS achieves a regret bound O(d √ K + d 2 ), where K is the number of episodes and d is the dimension of the feature mapping. As far as we know, HF-O 2 PS is the first algorithm for adversarial RL achieving a horizon-free regret upper bound.• We also provide hardness results in addition to the regret upper bound. Our first lower bound shows that an unbounded |S| will result in a lower bound asymptotically linear in √ H, which justifies our assumption of |S| < ∞. We also provide a minimax lower bound of Ω(d √ K), which manifests the near optimality of HF-O 2 PS.Notation We denote scalars by lowercase letters, and denote vectors and matrices by lower and uppercase boldface letters respectively. We use [n] to denote the set {1, . . . , n}, and [n] for set
|
Horizon-free Reinforcement Learning in Adversarial Linear Mixture MDPs
|
d264812826
|
Context-based fine-tuning methods, including prompting, in-context learning, soft prompting (also known as prompt tuning), and prefix-tuning, have gained popularity due to their ability to often match the performance of full fine-tuning with a fraction of the parameters.Despite their empirical successes, there is little theoretical understanding of how these techniques influence the internal computation of the model and their expressiveness limitations.We show that despite the continuous embedding space being more expressive than the discrete token space, soft-prompting and prefix-tuning are strictly less expressive than full fine-tuning, even with the same number of learnable parameters.Concretely, context-based fine-tuning cannot change the relative attention pattern over the content and can only bias the outputs of an attention layer in a fixed direction.This suggests that while techniques like prompting, in-context learning, soft prompting, and prefixtuning can effectively elicit skills present in the pretrained model, they cannot learn novel tasks that require new attention patterns.
|
WHEN DO PROMPTING AND PREFIX-TUNING WORK? A THEORY OF CAPABILITIES AND LIMITATIONS
|
d219558792
|
Efficient training of deep neural networks is an increasingly important problem in the era of sophisticated architectures and large-scale datasets. This paper proposes a training set synthesis technique, called Dataset Condensation, that learns to produce a small set of informative samples for training deep neural networks from scratch in a small fraction of the required computational cost on the original data while achieving comparable results. We rigorously evaluate its performance in several computer vision benchmarks and show that it significantly outperforms the state-of-the-art methods. Finally we show promising applications of our method in continual learning and domain adaptation.Preprint. Under review.
|
Dataset Condensation with Gradient Matching
|
d17220670
|
External neural memory structures have recently become a popular tool for algorithmic deep learning(Graves et al., 2014;Weston et al., 2014). These models generally utilize differentiable versions of traditional discrete memory-access structures (random access, stacks, tapes) to provide the storage necessary for computational tasks. In this work, we argue that these neural memory systems lack specific structure important for relative indexing, and propose an alternative model, Lieaccess memory, that is explicitly designed for the neural setting. In this paradigm, memory is accessed using a continuous head in a key-space manifold. The head is moved via Lie group actions, such as shifts or rotations, generated by a controller, and memory access is performed by linear smoothing in key space. We argue that Lie groups provide a natural generalization of discrete memory structures, such as Turing machines, as they provide inverse and identity operators while maintaining differentiability. To experiment with this approach, we implement a simplified Lie-access neural Turing machine (LANTM) with different Lie groups. We find that this approach is able to perform well on a range of algorithmic tasks.Recent work on neural Turing machines (NTMs)(Graves et al., 2014;and memory networks (MemNNs) (Weston et al., 2014) has repopularized the use of explicit external memory in neural networks and demonstrated that these networks can be effectively trained in an end-to-end fashion. These methods have been successfully applied to question answering (Weston et al.Grefenstette et al., 2015;Joulin & Mikolov, 2015), machine translation(Kalchbrenner et al., 2015), and other tasks. This methodology has the potential to extend deep networks in a general-purpose way beyond the limitations of fixed-length encodings such as standard recurrent neural networks (RNNs).A shared theme in many of these works (and earlier exploration of neural memory) is to re-frame traditional memory access paradigms to be continuous and possibly differentiable to allow for backpropagation. In MemNNs, traditional random-access memory is replaced with a ranking approach that finds the most likely memory. In the work ofGrefenstette et al. (2015), classical stack-, queue-, and deque-based memories are replaced by soft-differentiable stack, queue, and deque datastructures. In NTMs, sequential local-access memory is simulated by an explicit tape data structure.This work questions the assumption that neural memory should mimic the structure of traditional discrete memory. We argue that a neural memory should provide the following: (A) differentiability for end-to-end training and (B) robust relative indexing (perhaps in addition to random-access). Surprisingly many neural memory systems fail one of these conditions, either lacking Criterion B, discussed below, or employing extensions like REINFORCE to work around lack of differentiability(Zaremba & Sutskever, 2015).We propose instead a class of memory access techniques based around Lie groups, i.e. groups with differentiable operations, which provide a natural structure for neural memory access. By definition, their differentiability satisfies the concerns of Criterion A. Additionally the group axioms provide identity, invertibility, and associativity, all of which are desirable properties for a relative indexing scheme (Criterion B), and all of which are satisfied by standard Turing machines. Notably though, 1
|
LIE-ACCESS NEURAL TURING MACHINES
|
d257219732
|
We consider cross-silo federated linear contextual bandit (LCB) problem under differential privacy, where multiple silos (agents) interact with the local users and communicate via a central server to realize collaboration while without sacrificing each user's privacy. We identify three issues in the state-of-the-art: (i) failure of claimed privacy protection and (ii) incorrect regret bound due to noise miscalculation and (iii) ungrounded communication cost. To resolve these issues, we take a two-step principled approach. First, we design an algorithmic framework consisting of a generic federated LCB algorithm and flexible privacy protocols. Then, leveraging the proposed framework, we study federated LCBs under two different privacy constraints. We first establish privacy and regret guarantees under silo-level local differential privacy, which fix the issues present in state-of-the-art algorithm. To further improve the regret performance, we next consider shuffle model of differential privacy, under which we show that our algorithm can achieve nearly "optimal" regret without a trusted server. We accomplish this via two different schemes -one relies on a new result on privacy amplification via shuffling for DP mechanisms and another one leverages the integration of a shuffle protocol for vector sum into the tree-based mechanism, both of which might be of independent interest. Finally, we support our theoretical results with numerical evaluations over contextual bandit instances generated from both synthetic and real-life data.
|
On Differentially Private Federated Linear Contextual Bandits
|
d254854553
|
Robustness evaluation against adversarial examples has become increasingly important to unveil the trustworthiness of the prevailing deep models in natural language processing (NLP). However, in contrast to the computer vision (CV) domain where the first-order projected gradient descent (PGD) is used as the benchmark approach to generate adversarial examples for robustness evaluation, there lacks a principled first-order gradient-based robustness evaluation framework in NLP. The emerging optimization challenges lie in 1) the discrete nature of textual inputs together with the strong coupling between the perturbation location and the actual content, and 2) the additional constraint that the perturbed text should be fluent and achieve a low perplexity under a language model. These challenges make the development of PGD-like NLP attacks difficult. To bridge the gap, we propose TEXTGRAD, a new attack generator using gradient-driven optimization, supporting high-accuracy and high-quality assessment of adversarial robustness in NLP. Specifically, we address the aforementioned challenges in a unified optimization framework. And we develop an effective convex relaxation method to co-optimize the continuously-relaxed site selection and perturbation variables, and leverage an effective sampling method to establish an accurate mapping from the continuous optimization variables to the discrete textual perturbations. Moreover, as a first-order attack generation method, TEXTGRAD can be baked in adversarial training to further improve the robustness of NLP models. Extensive experiments are provided to demonstrate the effectiveness of TEXTGRAD not only in attack generation for robustness evaluation but also in adversarial defense. From the attack perspective, we show that TEXTGRAD achieves remarkable improvements in both the attack success rate and the perplexity score over five state-of-the-art baselines. From the defense perspective, TEXTGRAD-enabled adversarial training yields the most robust NLP model against a wide spectrum of NLP attacks. * Contributed equally.
|
TEXTGRAD: ADVANCING ROBUSTNESS EVALUATION IN NLP BY GRADIENT-DRIVEN OPTIMIZATION
|
d252531169
|
Extractive summarization produces summaries by identifying and concatenating the most important sentences in a document. Since most summarization datasets do not come with gold labels indicating whether document sentences are summary-worthy, different labeling algorithms have been proposed to extrapolate oracle extracts for model training. In this work, we identify two flaws with the widely used greedy labeling approach: it delivers suboptimal and deterministic oracles. To alleviate both issues, we propose a simple yet effective labeling algorithm that creates soft, expectation-based sentence labels. We define a new learning objective for extractive summarization which incorporates learning signals from multiple oracle summaries and prove it is equivalent to estimating the oracle expectation for each document sentence. Without any architectural modifications, the proposed labeling scheme achieves superior performance on a variety of summarization benchmarks across domains and languages, in both supervised and zero-shot settings. 1 1 Our code and models can be found at https://github.com/yumoxu/oreo.
|
TEXT SUMMARIZATION WITH ORACLE EXPECTATION
|
d245837268
|
Reward hacking-where RL agents exploit gaps in misspecified reward functions-has been widely observed, but not yet systematically studied. To understand how reward hacking arises, we construct four RL environments with misspecified rewards. We investigate reward hacking as a function of agent capabilities: model capacity, action space resolution, observation space noise, and training time. More capable agents often exploit reward misspecifications, achieving higher proxy reward and lower true reward than less capable agents. Moreover, we find instances of phase transitions: capability thresholds at which the agent's behavior qualitatively shifts, leading to a sharp decrease in the true reward. Such phase transitions pose challenges to monitoring the safety of ML systems. To address this, we propose an anomaly detection task for aberrant policies and offer several baseline detectors.
|
THE EFFECTS OF REWARD MISSPECIFICATION: MAPPING AND MITIGATING MISALIGNED MODELS
|
d257078985
|
In this paper, we propose energy-based sample adaptation at test time for domain generalization. Where previous works adapt their models to target domains, we adapt the unseen target samples to source-trained models. To this end, we design a discriminative energy-based model, which is trained on source domains to jointly model the conditional distribution for classification and data distribution for sample adaptation. The model is optimized to simultaneously learn a classifier and an energy function. To adapt target samples to source distributions, we iteratively update the samples by energy minimization with stochastic gradient Langevin dynamics. Moreover, to preserve the categorical information in the sample during adaptation, we introduce a categorical latent variable into the energy-based model. The latent variable is learned from the original sample before adaptation by variational inference and fixed as a condition to guide the sample update. Experiments on six benchmarks for classification of images and microblog threads demonstrate the effectiveness of our proposal. Brendel. A simple way to make neural networks robust against diverse image
|
ENERGY-BASED TEST SAMPLE ADAPTATION FOR DOMAIN GENERALIZATION
|
d20472740
|
Natural Language Inference (NLI) task requires an agent to determine the logical relationship between a natural language premise and a natural language hypothesis. We introduce Interactive Inference Network (IIN), a novel class of neural network architectures that is able to achieve high-level understanding of the sentence pair by hierarchically extracting semantic features from interaction space. We show that an interaction tensor (attention weight) contains semantic information to solve natural language inference, and a denser interaction tensor contains richer semantic information. One instance of such architecture, Densely Interactive Inference Network (DIIN), demonstrates the state-of-the-art performance on large scale NLI copora and large-scale NLI alike corpus. It's noteworthy that DIIN achieve a greater than 20% error reduction on the challenging Multi-Genre NLI (MultiNLI; Williams et al. 2017) dataset with respect to the strongest published system.
|
NATURAL LANGUAGE INFERENCE OVER INTERACTION SPACE
|
d260498700
|
Recurrent Neural Networks (RNN) are widely used to solve a variety of problems and as the quantity of data and the amount of available compute have increased, so have model sizes. The number of parameters in recent state-of-the-art networks makes them hard to deploy, especially on mobile phones and embedded devices. The challenge is due to both the size of the model and the time it takes to evaluate it. In order to deploy these RNNs efficiently, we propose a technique to reduce the parameters of a network by pruning weights during the initial training of the network. At the end of training, the parameters of the network are sparse while accuracy is still close to the original dense neural network. The network size is reduced by 8× and the time required to train the model remains constant. Additionally, we can prune a larger dense network to achieve better than baseline performance while still reducing the total number of parameters significantly. Pruning RNNs reduces the size of the model and can also help achieve significant inference time speed-up using sparse matrix multiply. Benchmarks show that using our technique model size can be reduced by 90% and speed-up is around 2× to 7×.Published as a conference paper at ICLR 2017The more powerful server class GPUs used in data centers can generally perform inference quickly enough to serve one user, but in the data center performance per dollar is very important. Techniques that allow models to be evaluated faster enable more users to be served per GPU increasing the effective performance per dollar.
|
EXPLORING SPARSITY IN RECURRENT NEURAL NETWORKS
|
d247594724
|
Differentiable sorting algorithms allow training with sorting and ranking supervision, where only the ordering or ranking of samples is known. Various methods have been proposed to address this challenge, ranging from optimal transport-based differentiable Sinkhorn sorting algorithms to making classic sorting networks differentiable. One problem of current differentiable sorting methods is that they are non-monotonic. To address this issue, we propose a novel relaxation of conditional swap operations that guarantees monotonicity in differentiable sorting networks. We introduce a family of sigmoid functions and prove that they produce differentiable sorting networks that are monotonic. Monotonicity ensures that the gradients always have the correct sign, which is an advantage in gradient-based optimization. We demonstrate that monotonic differentiable sorting networks improve upon previous differentiable sorting methods.
|
MONOTONIC DIFFERENTIABLE SORTING NETWORKS
|
d204734348
|
We propose a novel algorithm for learning fair representations that can simultaneously mitigate two notions of disparity among different demographic subgroups in the classification setting. Two key components underpinning the design of our algorithm are balanced error rate and conditional alignment of representations. We show how these two components contribute to ensuring accuracy parity and equalized false-positive and false-negative rates across groups without impacting demographic parity. Furthermore, we also demonstrate both in theory and on two real-world experiments that the proposed algorithm leads to a better utility-fairness trade-off on balanced datasets compared with existing algorithms on learning fair representations for classification.
|
CONDITIONAL LEARNING OF FAIR REPRESENTATIONS
|
d1859294
|
Recurrent Neural Networks (RNNs) continue to show outstanding performance in sequence modeling tasks. However, training RNNs on long sequences often face challenges like slow inference, vanishing gradients and difficulty in capturing long term dependencies. In backpropagation through time settings, these issues are tightly coupled with the large, sequential computational graph resulting from unfolding the RNN in time. We introduce the Skip RNN model which extends existing RNN models by learning to skip state updates and shortens the effective size of the computational graph. This model can also be encouraged to perform fewer state updates through a budget constraint. We evaluate the proposed model on various tasks and show how it can reduce the number of required RNN updates while preserving, and sometimes even improving, the performance of the baseline RNN models. Source code is publicly available at https://imatge-upc.github.io/skiprnn-2017-telecombcn/.
|
Skip RNN: Learning to Skip State Updates in Recurrent Neural Networks
|
d249461627
|
The performance of time series forecasting has recently been greatly improved by the introduction of transformers. In this paper, we propose a general multi-scale framework that can be applied to the state-of-the-art transformer-based time series forecasting models (FEDformer, Autoformer, etc.). By iteratively refining a forecasted time series at multiple scales with shared weights, introducing architecture adaptations, and a specially-designed normalization scheme, we are able to achieve significant performance improvements, from 5.5% to 38.5% across datasets and transformer architectures, with minimal additional computational overhead. Via detailed ablation studies, we demonstrate the effectiveness of each of our contributions across the architecture and methodology. Furthermore, our experiments on various public datasets demonstrate that the proposed improvements outperform their corresponding baseline counterparts. Our code is publicly available in https://github.com/BorealisAI/scaleformer.Published as a conference paper at ICLR 2023 forecasts which can lead to runaway error propagation. To mitigate this issue, we introduce cross-scale normalization at each step.Our approach re-orders model capacity to shift the focus on scale awareness, but does not fundamentally alter the attention-driven paradigm of transformers. As a result, it can be readily adapted to work jointly with multiple recent time series transformer architectures, acting broadly orthogonally to their own contributions. Leveraging this, we chose to operate with various transformer-based backbones (e.g. Fedformer, Autoformer, Informer, Reformer, Performer) to further probe the effect of our multi-scale method on a variety of experimental setups.Our contributions are as follows: (1) we introduce a novel iterative scale-refinement paradigm that can be readily adapted to a variety of transformer-based time series forecasting architectures. (2) To minimize distribution shifts between scales and windows, we introduce cross-scale normalization on outputs of the Transformer. (3) Using Informer and AutoFormer, two state-of-the-art architectures, as backbones, we demonstrate empirically the effectiveness of our approach on a variety of datasets. Depending on the choice of transformer architecture, our mutli-scale framework results in mean squared error reductions ranging from 5.5% to 38.5%. (4) Via a detailed ablation study of our findings, we demonstrate the validity of our architectural and methodological choices.
|
SCALEFORMER: ITERATIVE MULTI-SCALE REFINING TRANSFORMERS FOR TIME SERIES FORECASTING
|
d204837843
|
Graph Neural Networks (GNNs) are a powerful representational tool for solving problems on graph-structured inputs. In almost all cases so far, however, they have been applied to directly recovering a final solution from raw inputs, without explicit guidance on how to structure their problem-solving. Here, instead, we focus on learning in the space of algorithms: we train several state-of-the-art GNN architectures to imitate individual steps of classical graph algorithms, parallel (breadth-first search, Bellman-Ford) as well as sequential (Prim's algorithm). As graph algorithms usually rely on making discrete decisions within neighbourhoods, we hypothesise that maximisation-based message passing neural networks are bestsuited for such objectives, and validate this claim empirically. We also demonstrate how learning in the space of algorithms can yield new opportunities for positive transfer between tasks-showing how learning a shortest-path algorithm can be substantially improved when simultaneously learning a reachability algorithm. * Work performed while the author was at DeepMind. 1 We note that there exist good reasons for choosing this approach, e.g. ease of optimisation.Under review as a conference paper at ICLR 2020Given that the majority of popular algorithms requires making discrete decisions over neighbourhoods (e.g. "which edge should be taken?"), we suggest that a highly suitable architecture for this task is a message-passing neural network (Gilmer et al., 2017) with a maximisation aggregator-a claim we verify, demonstrating clear performance benefits for simultaneously learning breadth-first search for reachability with the Bellman-Ford algorithm for shortest paths. We also verify its applicability to sequential reasoning, through learning Prim's algorithm(Prim, 1957)for minimum spanning trees.Note that our approach complements Reed & De Freitas(2015): we show that a relatively simple graph neural network architecture is able to learn and algorithmically transfer among different tasks, do not require explicitly denoting subroutines, and tackle tasks with superlinear time complexity.
|
NEURAL EXECUTION OF GRAPH ALGORITHMS
|
d9655643
|
Sequence-to-sequence models rely on a fixed decomposition of the target sequences into a sequence of tokens that may be words, word-pieces or characters. The choice of these tokens and the decomposition of the target sequences into a sequence of tokens is often static, and independent of the input, output data domains. This can potentially lead to a sub-optimal choice of token dictionaries, as the decomposition is not informed by the particular problem being solved. In this paper we present Latent Sequence Decompositions (LSD), a framework in which the decomposition of sequences into constituent tokens is learnt during the training of the model. The decomposition depends both on the input sequence and on the output sequence. In LSD, during training, the model samples decompositions incrementally, from left to right by locally sampling between valid extensions. We experiment with the Wall Street Journal speech recognition task. Our LSD model achieves 12.9% WER compared to a character baseline of 14.8% WER. When combined with a convolutional network on the encoder, we achieve a WER of 9.6%. * Work done at Google Brain.
|
LATENT SEQUENCE DECOMPOSITIONS
|
d14992224
|
We introduce Deep Variational Bayes Filters (DVBF), a new method for unsupervised learning of latent Markovian state space models. Leveraging recent advances in Stochastic Gradient Variational Bayes, DVBF can overcome intractable inference distributions by means of variational inference. Thus, it can handle highly nonlinear input data with temporal and spatial dependencies such as image sequences without domain knowledge. Our experiments show that enabling backpropagation through transitions enforces state space assumptions and significantly improves information content of the latent embedding. This also enables realistic long-term prediction. * Justin Bayer is also affiliated with sensed.io UG (haftungsbeschränkt),
|
Deep Variational Bayes Filters: Unsupervised Learning of State Space Models from Raw Data
|
d8728609
|
Recent work has begun exploring neural acoustic word embeddings-fixeddimensional vector representations of arbitrary-length speech segments corresponding to words. Such embeddings are applicable to speech retrieval and recognition tasks, where reasoning about whole words may make it possible to avoid ambiguous sub-word representations. The main idea is to map acoustic sequences to fixed-dimensional vectors such that examples of the same word are mapped to similar vectors, while different-word examples are mapped to very different vectors. In this work we take a multi-view approach to learning acoustic word embeddings, in which we jointly learn to embed acoustic sequences and their corresponding character sequences. We use deep bidirectional LSTM embedding models and multi-view contrastive losses. We study the effect of different loss variants, including fixed-margin and cost-sensitive losses. Our acoustic word embeddings improve over previous approaches for the task of word discrimination. We also present results on other tasks that are enabled by the multi-view approach, including cross-view word discrimination and word similarity.One of the useful properties of this multi-view approach is that, unlike earlier work on acoustic word embeddings, it produces both acoustic and orthographic embeddings that can be directly compared. This makes it possible to use the same learned embeddings for multiple single-view and cross-view tasks. Our multi-view embeddings produce improved results over earlier work on acoustic word discrimination, as well as encouraging results on cross-view discrimination and word similarity. 1
|
MULTI-VIEW RECURRENT NEURAL ACOUSTIC WORD EMBEDDINGS
|
d264490854
|
Current tools for machine learning fairness only admit a limited range of fairness definitions and have seen little integration with automatic differentiation libraries, despite the central role these libraries play in modern machine learning pipelines.We introduce a framework of fairness regularization terms (FAIRRETs) which quantify bias as modular objectives that are easily integrated in automatic differentiation pipelines.By employing a general definition of fairness in terms of linear-fractional statistics, a wide class of FAIRRETs can be computed efficiently.Experiments show the behavior of their gradients and their utility in enforcing fairness with minimal loss of predictive power compared to baselines.Our contribution includes a PyTorch implementation of the FAIRRET framework.Related Work Fairness tools are classified as preprocessing, inprocessing or postprocessing(Mehrabi et al., 2021).FAIRRETs perform inprocessing, as they are minimized during training.
|
FAIRRET: A FRAMEWORK FOR DIFFERENTIABLE FAIRNESS REGULARIZATION TERMS
|
d231847016
|
We study the problem of how to construct a set of policies that can be composed together to solve a collection of reinforcement learning tasks. Each task is a different reward function defined as a linear combination of known features. We consider a specific class of policy compositions which we call set improving policies (SIPs): given a set of policies and a set of tasks, a SIP is any composition of the former whose performance is at least as good as that of its constituents across all the tasks. We focus on the most conservative instantiation of SIPs, setmax policies (SMPs), so our analysis extends to any SIP. This includes known policy-composition operators like generalized policy improvement. Our main contribution is a policy iteration algorithm that builds a set of policies in order to maximize the worst-case performance of the resulting SMP on the set of tasks. The algorithm works by successively adding new policies to the set. We show that the worst-case performance of the resulting SMP strictly improves at each iteration, and the algorithm only stops when there does not exist a policy that leads to improved performance. We empirically evaluate our algorithm on a grid world and also on a set of domains from the DeepMind control suite. We confirm our theoretical results regarding the monotonically improving performance of our algorithm. Interestingly, we also show empirically that the sets of policies computed by the algorithm are diverse, leading to different trajectories in the grid world and very distinct locomotion skills in the control suite. . Fast reinforcement learning with generalized policy updates. . Diversity is all you need: Learning skills without a reward function. arXiv preprint arXiv:1802.06070, 2018. Marguerite Frank and Philip Wolfe. An algorithm for quadratic programming. Naval research logistics quarterly, 3(1-2):95-110, 1956. Dan Garber and Elad Hazan. A linearly convergent conditional gradient algorithm with applications to online and stochastic optimization. arXiv preprint arXiv:1301.4666, 2013. Reinforcement learning with soft state aggregation. In Advances in neural information processing systems, pp. 361-368, 1995. Nathan Sprague and Dana Ballard. Multiple-goal reinforcement learning with modular sarsa (0). 2003.
|
DISCOVERING A SET OF POLICIES FOR THE WORST CASE REWARD
|
d49907212
|
We study the problem of generating adversarial examples in a black-box setting in which only lossoracle access to a model is available. We introduce a framework that conceptually unifies much of the existing work on black-box attacks, and we demonstrate that the current state-of-the-art methods are optimal in a natural sense. Despite this optimality, we show how to improve black-box attacks by bringing a new element into the problem: gradient priors. We give a bandit optimization-based algorithm that allows us to seamlessly integrate any such priors, and we explicitly identify and incorporate two examples. The resulting methods use two to four times fewer queries and fail two to five times less often than the current state-of-the-art. 1 * Equal contribution 1 The code for reproducing our work is available at https://git.io/blackbox-bandits. arXiv:1807.07978v3 [stat.ML] to break through the optimality of current methods. We evaluate our approach on the task of generating black-box adversarial examples, where the methods obtained from integrating two example priors significantly outperform state-of-the-art approaches.Concretely, in this work:1. We formalize the gradient estimation problem as the central problem in the context of query-efficient black-box attacks. We then show how the resulting framework unifies the previous attack methodology. We prove that the least squares method, a classic primitive in signal processing, not only constitutes an optimal solution to the general gradient estimation problem but also is essentially equivalent to the current-best black-box attack methods.2. We demonstrate that, despite this seeming optimality of these methods, we can still improve upon them by exploiting an aspect of the problem that has been not considered previously: the priors we have on the distribution of the gradient. We identify two example classes of such priors, and show that they indeed lead to better predictors of the gradient.3. Finally, we develop a bandit optimization framework for generating black-box adversarial examples which allows for the seamless integration of priors. To demonstrate its effectiveness, we show that leveraging the two aforementioned priors yields black-box attacks that are 2-5 times more query efficient and less failure-prone than the state of the art.
|
Prior Convictions: Black-Box Adversarial Attacks with Bandits and Priors
|
d221655222
|
In this work, we propose information laundering, a novel framework for enhancing model privacy. Unlike data privacy that concerns the protection of raw data information, model privacy aims to protect an already-learned model that is to be deployed for public use. The private model can be obtained from general learning methods, and its deployment means that it will return a deterministic or random response for a given input query. An informationlaundered model consists of probabilistic components that deliberately maneuver the intended input and output for queries to the model, so the model's adversarial acquisition is less likely. Under the proposed framework, we develop an information-theoretic principle to quantify the fundamental tradeoffs between model utility and privacy leakage and derive the optimal design.
|
Information Laundering for Model Privacy
|
d263310924
|
Seminal research in the field of graph neural networks (GNNs) has revealed a direct correspondence between the expressive capabilities of GNNs and the kdimensional Weisfeiler-Leman (kWL) test, a widely-recognized method for verifying graph isomorphism. This connection has reignited interest in comprehending the specific graph properties effectively distinguishable by the kWL test. A central focus of research in this field revolves around determining the least dimensionality k, for which kWL can discern graphs with different number of occurrences of a pattern graph P . We refer to such a least k as the WL-dimension of this pattern counting problem. This inquiry traditionally delves into two distinct counting problems related to patterns: subgraph counting and induced subgraph counting. Intriguingly, despite their initial appearance as separate challenges with seemingly divergent approaches, both of these problems are interconnected components of a more comprehensive problem: "graph motif parameters". In this paper, we provide a precise characterization of the WL-dimension of labeled graph motif parameters. As specific instances of this result, we obtain characterizations of the WL-dimension of the subgraph counting and induced subgraph counting problem for every labeled pattern P . Particularly noteworthy is our resolution of a problem left open in previous work concerning induced copies. We additionally demonstrate that in cases where the kWL test distinguishes between graphs with varying occurrences of a pattern P , the exact number of occurrences of P can be computed uniformly using only local information of the last layer of a corresponding GNN. We finally delve into the challenge of recognizing the WLdimension of various graph parameters. We give a polynomial time algorithm for determining the WL-dimension of the subgraph counting problem for given pattern P , answering an open question from previous work. We additionally show how to utilize deep results from the field of graph motif parameters, together with our characterization, to determine the WL-dimension of induced subgraph counting and counting k-graphlets. cally at http://oeis.org.Victor M. Preciado and AliJadbabaie. From local measurements to network spectral properties: Beyond degree distributions. In CDC, pp. 2686-2691. IEEE, 2010. David E. Roberson. Oddomorphisms and homomorphism indistinguishability over graphs of bounded degree. CoRR, abs/2206.10321, 2022. . Counting induced subgraphs: A topological approach to #w[1]hardness. Algorithmica, 82(8):2267-2291, 2020.
|
ON THE POWER OF THE WEISFEILER-LEMAN TEST FOR GRAPH MOTIF PARAMETERS
|
d52895589
|
Graph Neural Networks (GNNs) for representation learning of graphs broadly follow a neighborhood aggregation framework, where the representation vector of a node is computed by recursively aggregating and transforming feature vectors of its neighboring nodes. Many GNN variants have been proposed and have achieved state-of-the-art results on both node and graph classification tasks. However, despite GNNs revolutionizing graph representation learning, there is limited understanding of their representational properties and limitations. Here, we present a theoretical framework for analyzing the expressive power of GNNs in capturing different graph structures. Our results characterize the discriminative power of popular GNN variants, such as Graph Convolutional Networks and GraphSAGE, and show that they cannot learn to distinguish certain simple graph structures. We then develop a simple architecture that is provably the most expressive among the class of GNNs and is as powerful as the Weisfeiler-Lehman graph isomorphism test. We empirically validate our theoretical findings on a number of graph classification benchmarks, and demonstrate that our model achieves state-of-the-art performance.
|
HOW POWERFUL ARE GRAPH NEURAL NETWORKS?
|
d222379753
|
All few-shot learning techniques must be pre-trained on a large, labeled "base dataset". In problem domains where such large labeled datasets are not available for pre-training (e.g., X-ray images), one must resort to pre-training in a different "source" problem domain (e.g., ImageNet), which can be very different from the desired target task. Traditional few-shot and transfer learning techniques fail in the presence of such extreme differences between the source and target tasks. In this paper, we present a simple and effective solution to tackle this extreme domain gap: self-training a source domain representation on unlabeled data from the target domain. We show that this improves one-shot performance on the target domain by 2.9 points on average on a challenging benchmark with multiple domains.
|
SELF-TRAINING FOR FEW-SHOT TRANSFER ACROSS EXTREME TASK DIFFERENCES
|
d246210276
|
Backdoor attacks (BAs) are an emerging threat to deep neural network classifiers. A victim classifier will predict to an attacker-desired target class whenever a test sample is embedded with the same backdoor pattern (BP) that was used to poison the classifier's training set. Detecting whether a classifier is backdoor attacked is not easy in practice, especially when the defender is, e.g., a downstream user without access to the classifier's training set. This challenge is addressed here by a reverse-engineering defense (RED), which has been shown to yield state-of-the-art performance in several domains. However, existing REDs are not applicable when there are only two classes or when multiple attacks are present. These scenarios are first studied in the current paper, under the practical constraints that the defender neither has access to the classifier's training set nor to supervision from clean reference classifiers trained for the same domain. We propose a detection framework based on BP reverseengineering and a novel expected transferability (ET) statistic. We show that our ET statistic is effective using the same detection threshold, irrespective of the classification domain, the attack configuration, and the BP reverse-engineering algorithm that is used. The excellent performance of our method is demonstrated on six benchmark datasets. Notably, our detection framework is also applicable to multi-class scenarios with multiple attacks. Code is available at
|
POST-TRAINING DETECTION OF BACKDOOR ATTACKS FOR TWO-CLASS AND MULTI-ATTACK SCENARIOS
|
d263334074
|
Despite the success of large language models (LLMs), the task of theorem proving still remains one of the hardest reasoning tasks that is far from being fully solved.Prior methods using language models have demonstrated promising results, but they still struggle to prove even middle school level theorems.One common limitation of these methods is that they assume a fixed theorem library during the whole theorem proving process.However, as we all know, creating new useful theorems or even new theories is not only helpful but crucial and necessary for advancing mathematics and proving harder and deeper results.In this work, we present LEGO-Prover, which employs a growing skill library containing verified lemmas as skills to augment the capability of LLMs used in theorem proving.By constructing the proof modularly, LEGO-Prover enables LLMs to utilize existing skills retrieved from the library and to create new skills during the proving process.These skills are further evolved (by prompting an LLM) to enrich the library on another scale.Modular and reusable skills are constantly added to the library to enable tackling increasingly intricate mathematical problems.Moreover, the learned library further bridges the gap between human proofs and formal proofs by making it easier to impute missing steps.LEGO-Prover advances the stateof-the-art pass rate on miniF2F-valid (48.0% to 57.0%) and miniF2F-test (45.5% to 50.0%).During the proving process, LEGO-Prover also manages to generate over 20,000 skills (theorems/lemmas) and adds them to the growing library.Our ablation study indicates that these newly added skills are indeed helpful for proving theorems, resulting in an improvement from a success rate of 47.1% to 50.4%.We also release our code and all the generated skills. 1
|
LEGO-PROVER: NEURAL THEOREM PROVING WITH GROWING LIBRARIES
|
d238198403
|
Voice conversion is a common speech synthesis task which can be solved in different ways depending on a particular real-world scenario. The most challenging one often referred to as one-shot many-to-many voice conversion consists in copying target voice from only one reference utterance in the most general case when both source and target speakers do not belong to the training dataset. We present a scalable high-quality solution based on diffusion probabilistic modeling and demonstrate its superior quality compared to state-of-the-art one-shot voice conversion approaches. Moreover, focusing on real-time applications, we investigate general principles which can make diffusion models faster while keeping synthesis quality at a high level. As a result, we develop a novel Stochastic Differential Equations solver suitable for various diffusion model types and generative tasks as shown through empirical studies and justify it by theoretical analysis. The code is publicly available at https://github.com/huawei-noah/ Speech-Backbones/tree/main/DiffVC.
|
DIFFUSION-BASED VOICE CONVERSION WITH FAST MAXIMUM LIKELIHOOD SAMPLING SCHEME
|
d52900202
|
Brain-Machine Interfaces (BMIs) have recently emerged as a clinically viable option to restore voluntary movements after paralysis. These devices are based on the ability to extract information about movement intent from neural signals recorded using multi-electrode arrays chronically implanted in the motor cortices of the brain. However, the inherent loss and turnover of recorded neurons requires repeated recalibrations of the interface, which can potentially alter the day-to-day user experience. The resulting need for continued user adaptation interferes with the natural, subconscious use of the BMI. Here, we introduce a new computational approach that decodes movement intent from a low-dimensional latent representation of the neural data. We implement various domain adaptation methods to stabilize the interface over significantly long times. This includes Canonical Correlation Analysis used to align the latent variables across days; this method requires prior point-to-point correspondence of the time series across domains. Alternatively, we match the empirical probability distributions of the latent variables across days through the minimization of their Kullback-Leibler divergence. These two methods provide a significant and comparable improvement in the performance of the interface. However, implementation of an Adversarial Domain Adaptation Network trained to match the empirical probability distribution of the residuals of the reconstructed neural signals outperforms the two methods based on latent variables, while requiring remarkably few data points to solve the domain adaptation problem.
|
ADVERSARIAL DOMAIN ADAPTATION FOR STABLE BRAIN-MACHINE INTERFACES
|
d60441438
|
Intelligent agents can learn to represent the action spaces of other agents simply by observing them act.Such representations help agents quickly learn to predict the effects of their own actions on the environment and to plan complex action sequences.In this work, we address the problem of learning an agent's action space purely from visual observation.We use stochastic video prediction to learn a latent variable that captures the scene's dynamics while being minimally sensitive to the scene's static content.We introduce a loss term that encourages the network to capture the composability of visual sequences and show that it leads to representations that disentangle the structure of actions.We call the full model with composable action representations Composable Learned Action Space Predictor (CLASP) .We show the applicability of our method to synthetic settings and its potential to capture action spaces in complex, realistic visual settings.When used in a semi-supervised setting, our learned representations perform comparably to existing fully supervised methods on tasks such as action-conditioned video prediction and planning in the learned action space, while requiring orders of magnitude fewer action labels. 1 * Equal contribution.Ordering determined by a coin flip. 1 Project website: https://daniilidis-group.github.io/learned_action_spaces
|
LEARNING WHAT YOU CAN DO BEFORE DOING ANYTHING
|
d219572891
|
Learned Bloom filters enhance standard Bloom filters by using a learned model for the represented data set. However, a learned Bloom filter may under-utilize the model by not taking full advantage of the output. The learned Bloom filter uses the output score by simply applying a threshold, with elements above the threshold being interpreted as positives, and elements below the threshold subject to further analysis independent of the output score (using a smaller backup Bloom filter to prevent false negatives). While recent work has suggested additional heuristic approaches to take better advantage of the score, the results are only heuristic. Here, we instead frame the problem of optimal model utilization as an optimization problem. We show that the optimization problem can be effectively solved efficiently, yielding an improved partitioned learned Bloom filter, which partitions the score space and utilizes separate backup Bloom filters for each region. Experimental results from both simulated and real-world datasets show significant performance improvements from our optimization approach over both the original learned Bloom filter constructions and previously proposed heuristic improvements.
|
Partitioned Learned Bloom Filters
|
d264490454
|
To correctly use in-context information, language models (LMs) must bind entities to their attributes.For example, given a context describing a "green square" and a "blue circle", LMs must bind the shapes to their respective colors.We analyze LM representations and identify the binding ID mechanism: a general mechanism for solving the binding problem, which we observe in every sufficiently large model from the Pythia and LLaMA families.Using causal interventions, we show that LMs' internal activations represent binding information by attaching binding ID vectors to corresponding entities and attributes.We further show that binding ID vectors form a continuous subspace, in which distances between binding ID vectors reflect their discernability.Overall, our results uncover interpretable strategies in LMs for representing symbolic knowledge in-context, providing a step towards understanding general in-context reasoning in large-scale LMs.
|
HOW DO LANGUAGE MODELS BIND ENTITIES IN CONTEXT?
|
d52893515
|
We investigate the loss surface of neural networks. We prove that even for one-hidden-layer networks with "slightest" nonlinearity, the empirical risks have spurious local minima in most cases. Our results thus indicate that in general "no spurious local minima" is a property limited to deep linear networks, and insights obtained from linear networks are not robust. Specifically, for ReLU(-like) networks we constructively prove that for almost all (in contrast to previous results) practical datasets there exist infinitely many local minima. We also present a counterexample for more general activations (sigmoid, tanh, arctan, ReLU, etc.), for which there exists a bad local minimum. Our results make the least restrictive assumptions relative to existing results on local optimality in neural networks. We complete our discussion by presenting a comprehensive characterization of global optimality for deep linear networks, which unifies other results on this topic.
|
Small nonlinearities in activation functions create bad local minima in neural networks
|
d51926976
|
The ability to generate natural language sequences from source code snippets has a variety of applications such as code summarization, documentation, and retrieval. Sequence-to-sequence (seq2seq) models, adopted from neural machine translation (NMT), have achieved state-of-the-art performance on these tasks by treating source code as a sequence of tokens. We present CODE2SEQ: an alternative approach that leverages the syntactic structure of programming languages to better encode source code. Our model represents a code snippet as the set of compositional paths in its abstract syntax tree (AST) and uses attention to select the relevant paths while decoding. We demonstrate the effectiveness of our approach for two tasks, two programming languages, and four datasets of up to 16M examples. Our model significantly outperforms previous models that were specifically designed for programming languages, as well as state-of-the-art NMT models.
|
CODE2SEQ: GENERATING SEQUENCES FROM STRUCTURED REPRESENTATIONS OF CODE
|
d209977508
|
This paper introduces Meta-Q-Learning (MQL), a new off-policy algorithm for meta-Reinforcement Learning (meta-RL). MQL builds upon three simple ideas. First, we show that Q-learning is competitive with state-of-the-art meta-RL algorithms if given access to a context variable that is a representation of the past trajectory. Second, a multi-task objective to maximize the average reward across the training tasks is an effective method to meta-train RL policies. Third, past data from the meta-training replay buffer can be recycled to adapt the policy on a new task using off-policy updates. MQL draws upon ideas in propensity estimation to do so and thereby amplifies the amount of available data for adaptation. Experiments on standard continuous-control benchmarks suggest that MQL compares favorably with the state of the art in meta-RL.
|
META-Q-LEARNING
|
d65455367
|
Several recently proposed stochastic optimization methods that have been successfully used in training deep networks such as RMSPROP, ADAM, ADADELTA, NADAM are based on using gradient updates scaled by square roots of exponential moving averages of squared past gradients. In many applications, e.g. learning with large output spaces, it has been empirically observed that these algorithms fail to converge to an optimal solution (or a critical point in nonconvex settings). We show that one cause for such failures is the exponential moving average used in the algorithms. We provide an explicit example of a simple convex optimization setting where ADAM does not converge to the optimal solution, and describe the precise problems with the previous analysis of ADAM algorithm. Our analysis suggests that the convergence issues can be fixed by endowing such algorithms with "long-term memory" of past gradients, and propose new variants of the ADAM algorithm which not only fix the convergence issues but often also lead to improved empirical performance.
|
ON THE CONVERGENCE OF ADAM AND BEYOND
|
d3300406
|
Options in reinforcement learning allow agents to hierarchically decompose a task into subtasks, having the potential to speed up learning and planning. However, autonomously learning effective sets of options is still a major challenge in the field. In this paper we focus on the recently introduced idea of using representation learning methods to guide the option discovery process. Specifically, we look at eigenoptions, options obtained from representations that encode diffusive information flow in the environment. We extend the existing algorithms for eigenoption discovery to settings with stochastic transitions and in which handcrafted features are not available. We propose an algorithm that discovers eigenoptions while learning non-linear state representations from raw pixels. It exploits recent successes in the deep reinforcement learning literature and the equivalence between proto-value functions and the successor representation. We use traditional tabular domains to provide intuition about our approach and Atari 2600 games to demonstrate its potential.
|
EIGENOPTION DISCOVERY THROUGH THE DEEP SUCCESSOR REPRESENTATION
|
d219956317
|
Modern large-scale machine learning applications require stochastic optimization algorithms to be implemented on distributed compute systems. A key bottleneck of such systems is the communication overhead for exchanging information across the workers, such as stochastic gradients. Among the many techniques proposed to remedy this issue, one of the most successful is the framework of compressed communication with error feedback (EF). EF remains the only known technique that can deal with the error induced by contractive compressors which are not unbiased, such as Top-K. In this paper, we propose a new and theoretically and practically better alternative to EF for dealing with contractive compressors. In particular, we propose a construction which can transform any contractive compressor into an induced unbiased compressor. Following this transformation, existing methods able to work with unbiased compressors can be applied. We show that our approach leads to vast improvements over EF, including reduced memory requirements, better communication complexity guarantees and fewer assumptions. We further extend our results to federated learning with partial participation following an arbitrary distribution over the nodes, and demonstrate the benefits thereof. We perform several numerical experiments which validate our theoretical findings.
|
A Better Alternative to Error Feedback for Communication-Efficient Distributed Learning
|
d3720457
|
Careful tuning of the learning rate, or even schedules thereof, can be crucial to effective neural net training. There has been much recent interest in gradient-based meta-optimization, where one tunes hyperparameters, or even learns an optimizer, in order to minimize the expected loss when the training procedure is unrolled. But because the training procedure must be unrolled thousands of times, the metaobjective must be defined with an orders-of-magnitude shorter time horizon than is typical for neural net training. We show that such short-horizon meta-objectives cause a serious bias towards small step sizes, an effect we term short-horizon bias. We introduce a toy problem, a noisy quadratic cost function, on which we analyze short-horizon bias by deriving and comparing the optimal schedules for short and long time horizons. We then run meta-optimization experiments (both offline and online) on standard benchmark datasets, showing that meta-optimization chooses too small a learning rate by multiple orders of magnitude, even when run with a moderately long time horizon (100 steps) typical of work in the area. We believe short-horizon bias is a fundamental problem that needs to be addressed if metaoptimization is to scale to practical neural net training regimes. * Equal contribution.
|
UNDERSTANDING SHORT-HORIZON BIAS IN STOCHASTIC META-OPTIMIZATION
|
d263831046
|
We tackle the challenge of large-scale network intervention for guiding excitatory point processes, such as infectious disease spread or traffic congestion control.Our model-based reinforcement learning utilizes neural ODEs to capture how the networked excitatory point processes will evolve subject to the time-varying changes in network topology.Our approach incorporates Gradient-Descent based Model Predictive Control (GD-MPC), offering policy flexibility to accommodate prior knowledge and constraints.To address the intricacies of planning and overcome the high dimensionality inherent to such decision-making problems, we design an Amortize Network Interventions (ANI) framework, allowing for the pooling of optimal policies from history and other contexts, while ensuring a permutation equivalent property.This property enables efficient knowledge transfer and sharing across diverse contexts.Our approach has broad applications, from curbing infectious disease spread to reducing carbon emissions through traffic light optimization, and thus has the potential to address critical societal and environmental challenges.
|
AMORTIZED NETWORK INTERVENTION TO STEER THE EXCITATORY POINT PROCESSES
|
d257405190
|
Figure 1. Our model supports photo-realistic large-hole inpainting for various scenarios. The first example for object removal is a highresolution image captured in the wild, while other inpainting examples (512 × 512) come from Places2 [82] and CelebA-HQ [23] datasets.AbstractGenerative adversarial networks (GANs) have made great success in image inpainting yet still have difficulties tackling large missing regions. In contrast, iterative probabilistic algorithms, such as autoregressive and denoising diffusion models, have to be deployed with massive computing resources for decent effect. To achieve high-quality results with low computational cost, we present a novel pixel spread model (PSM) that iteratively employs decoupled probabilistic modeling, combining the optimization efficiency of GANs with the prediction tractability of probabilistic models. As a result, our model selectively spreads informative pixels throughout the image in a few iterations, largely enhancing the completion quality and efficiency. On multiple benchmarks, we achieve new state-of-the-art performance. Code is released at https://github.com/ fenglinglwb/PSM .
|
Image Inpainting via Iteratively Decoupled Probabilistic Modeling
|
d3502468
|
Incremental class learning involves sequentially learning classes in bursts of examples from the same class. This violates the assumptions that underlie methods for training standard deep neural networks, and will cause them to suffer from catastrophic forgetting. Arguably, the best method for incremental class learning is iCaRL, but it requires storing training examples for each class, making it challenging to scale. Here, we propose FearNet for incremental class learning. Fear-Net is a generative model that does not store previous examples, making it memory efficient. FearNet uses a brain-inspired dual-memory system in which new memories are consolidated from a network for recent memories inspired by the mammalian hippocampal complex to a network for long-term storage inspired by medial prefrontal cortex. Memory consolidation is inspired by mechanisms that occur during sleep. FearNet also uses a module inspired by the basolateral amygdala for determining which memory system to use for recall. FearNet achieves state-of-the-art performance at incremental class learning on image (CIFAR-100, CUB-200) and audio classification (AudioSet) benchmarks.
|
FEARNET: BRAIN-INSPIRED MODEL FOR INCREMENTAL LEARNING
|
d235899205
|
We introduce HTLM, a hyper-text language model trained on a large-scale web crawl. Modeling hyper-text has a number of advantages: (1) it is easily gathered at scale, (2) it provides rich document-level and end-taskadjacent supervision (e.g. class and id attributes often encode document category information), and (3) it allows for new structured prompting that follows the established semantics of HTML (e.g. to do zero-shot summarization by infilling <title> tags for a webpage that contains the input text). We show that pretraining with a BART-style denoising loss directly on simplified HTML provides highly effective transfer for a wide range of end tasks and supervision levels. HTLM matches or exceeds the performance of comparably sized text-only LMs for zero-shot prompting and fine-tuning for classification benchmarks, while also setting new state-of-the-art performance levels for zero-shot summarization. We also find that hyper-text prompts provide more value to HTLM, in terms of data efficiency, than plain text prompts do for existing LMs, and that HTLM is highly effective at autoprompting itself, by simply generating the most likely hyper-text formatting for any available training data. We will release all code and models to support future HTLM research.
|
HTLM: Hyper-Text Pre-Training and Prompting of Language Models
|
d252668463
|
Explaining generalizations and preventing over-confident predictions are central goals of studies on the loss landscape of neural networks. Flatness, defined as loss invariability on perturbations of a pre-trained solution, is widely accepted as a predictor of generalization in this context. However, the problem that flatness and generalization bounds can be changed arbitrarily according to the scale of a parameter was pointed out, and previous studies partially solved the problem with restrictions: Counter-intuitively, their generalization bounds were still variant for the functionpreserving parameter scaling transformation or limited only to an impractical network structure. As a more fundamental solution, we propose new prior and posterior distributions invariant to scaling transformations by decomposing the scale and connectivity of parameters, thereby allowing the resulting generalization bound to describe the generalizability of a broad class of networks with the more practical class of transformations such as weight decay with batch normalization. We also show that the above issue adversely affects the uncertainty calibration of Laplace approximation and propose a solution using our invariant posterior. We empirically demonstrate our posterior provides effective flatness and calibration measures with low complexity in such a practical parameter transformation case, supporting its practical effectiveness in line with our rationale. * Co-corresponding author arXiv:2209.15208v1 [cs.LG] 30 Sep 2022 Nonetheless, the limitations of the FM hypothesis were pointed out by Dinh et al. [10], Li et al. [11]. By rescaling two successive layers, Dinh et al.[10] demonstrated it was possible to modify a flatness measure without modifying the functions, hence allowing extraneous variability to be captured in the computation of generalizability. Meanwhile, Li et al.[11]argued that weight decayregularization[12]is an important limitation of the FM hypothesis as it leads to a contradiction of the FM hypothesis in practice; the weight decay sharpens the pre-trained solutions of NNs by downscaling the parameters, whereas the weight decay actually improves the generalization performance of NNs in general cases[13]. In short, they suggest that scaling transformation on network parameters (e.g., re-scaling layers and weight decay) may lead to a contradiction of the FM hypothesis.To resolve this contradiction, we investigate PAC-Bayesian prior and posterior distributions to derive a new scale-invariant generalization bound. Unlike related works[14,15], our bound guarantees the invariance for a general class of function-preserving parameter scaling transformation with a broad class of networks [16] (Secs.2.2 and 2.3).This bound is derived from the scale invariance of the prior and poster distributions, which guarantees not only the scale invariance of the bound but also its substance the Kullback-Leibler (KL) divergencebased kernel; we named this new term with scale-invariance property as empirical Connectivity Tangent Kernel (CTK) as it can be considered as a modification of empirical Neural Tangent Kernel[17]. Consequently, we define a novel sharpness metric named Connectivity Sharpness (CS) as a trace of CTK. Empirically, we verify our CS has a better prediction for generalization performance of NNs than existing sharpness measures[18,19,20]with a low-complexity (Sec. 2.5), with confirming its stronger correlation to generalization error (Sec. 4.1).We also found the contradiction of the FM hypothesis turns into meaningless predictive uncertainty amplifying issues in the Bayesian NN regime (Sec. 3.1), and can alleviate this issue by using Bayesian NN based on the posterior distribution of our PAC-Bayesian analysis. We call the resulting Bayesian NN as Connectivity Laplace (CL), as it can be seen as a variation of Laplace Approximation (LA; MacKay [8]) using different Jacobian. In particular, we provide pitfalls of weight decay regularization with BN in LA and its remedy using our posterior (Sec. 3.1) to suggest practical utilities of our Bayesian NNs (Sec. 4.2). We summarize our contributions as follows:
|
Scale-invariant Bayesian Neural Networks with Connectivity Tangent Kernel
|
d173990564
|
Neural Networks (NNs) have been extensively used for a wide spectrum of realworld regression tasks, where the goal is to predict a numerical outcome such as revenue, effectiveness, or a quantitative result. In many such tasks, the point prediction is not enough, but also the uncertainty (i.e. risk, or confidence) of that prediction must be estimated. Standard NNs, which are most often used in such tasks, do not provide any such information. Existing approaches try to solve this issue by combining Bayesian models with NNs, but these models are hard to implement, more expensive to train, and usually do not perform as well as standard NNs. In this paper, a new framework called RIO is developed that makes it possible to estimate uncertainty in any pretrained standard NN. RIO models prediction residuals using Gaussian Process with a composite input/output kernel. The residual prediction and I/O kernel are theoretically motivated and the framework is evaluated in twelve real-world datasets. It is found to provide reliable estimates of the uncertainty, reduce the error of the point predictions, and scale well to large datasets. Given that RIO can be applied to any standard NN without modifications to model architecture or training pipeline, it provides an important ingredient in building real-world applications of NNs.Preprint. Under review.
|
Quantifying Point-Prediction Uncertainty in Neural Networks via Residual Estimation with an I/O Kernel
|
d203734686
|
The critical locus of the loss function of a neural network is determined by the geometry of the functional space and by the parameterization of this space by the network's weights. We introduce a natural distinction between pure critical points, which only depend on the functional space, and spurious critical points, which arise from the parameterization. We apply this perspective to revisit and extend the literature on the loss function of linear neural networks. For this type of network, the functional space is either the set of all linear maps from input to output space, or a determinantal variety, i.e., a set of linear maps with bounded rank. We use geometric properties of determinantal varieties to derive new results on the landscape of linear networks with different loss functions and different parameterizations. * Equal contribution.In particular, we prove that non-global local minima are necessarily pure critical points for convex losses, which means that many properties of the loss landscape can be read from the functional space. On the other hand, we emphasize that even for linear networks it is possible to find many smooth convex losses with non-global local minima. This happens when the functional space is a determinantal variety, i.e., a (non-smooth and non-convex) family of matrices with bounded rank. In this setting, the absence of non-global minima actually holds in the particular case of the quadratic loss, because of very special geometric properties of determinantal varieties that we discuss. showed that linear networks with "no bottlenecks" have no bad local minima for arbitrary smooth loss functions. Lu & Kawaguchi (2017) and Zhang (2019) argued that "depth does not create local minima", meaning that the absence of local minima of deep linear networks is implied by the same property of shallow linear networks. Our study of pure and spurious critical points can be used as a framework for explaining all these results in a unified and intuitive way. Our analysis is also closely related to objects of study in applied algebraic geometry, particularly determinantal varieties and ED discriminants(Draisma et al., 2013;Ottaviani et al., 2013).Main contributions.• We introduce a natural distinction between "pure" and "spurious" critical points for the loss function of networks. While most of the paper focuses on linear networks, this viewpoint applies to more general settings as well (see also our discussion in Appendix A.3).• We study the pure and critical locus for linear networks and arbitrary loss functions. We show that non-global local minima are always pure for convex losses, unifying many known properties on the landscape of linear networks. We also prove other new results, including a precise description of the number of topologically connected components of the set of global minima.
|
PURE AND SPURIOUS CRITICAL POINTS: A GEOMETRIC STUDY OF LINEAR NETWORKS
|
d236170938
|
Learning the structure of a causal graphical model using both observational and interventional data is a fundamental problem in many scientific fields. A promising direction is continuous optimization for score-based methods, which, however, require constrained optimization to enforce acyclicity or lack convergence guarantees. In this paper, we present ENCO, an efficient structure learning method for directed, acyclic causal graphs leveraging observational and interventional data. ENCO formulates the graph search as an optimization of independent edge likelihoods, with the edge orientation being modeled as a separate parameter. Consequently, we provide for ENCO convergence guarantees when interventions on all variables are available, without having to constrain the score function with respect to acyclicity. In experiments, we show that ENCO can efficiently recover graphs with hundreds of nodes, an order of magnitude larger than what was previously possible, while handling deterministic variables and discovering latent confounders. * Qualcomm AI Research is an initiative of Qualcomm Technologies, Inc.In this work, we address both problems. By modeling the orientation of an edge as a separate parameter, we can define the score function without any acyclicity constraints or regularizers. This allows for unbiased low-variance gradient estimators that scale learning to much larger graphs. Yet, if we are able to intervene on all variables, the proposed optimization is guaranteed to converge to the correct, acyclic graph. Importantly, since such interventions might not always be available, we show that our algorithm performs robustly even when intervening on fewer variables and having small sample sizes. We call our method ENCO for Efficient Neural Causal Discovery.We make the following four contributions. Firstly, we propose ENCO, a causal structure learning method for observational and interventional data using continuous optimization. Different from recent methods, ENCO models the edge orientation as a separate parameter. Secondly, we derive unbiased, low-variance gradient estimators, which is crucial for scaling up the model to large numbers of variables. Thirdly, we show that ENCO is guaranteed to converge to the correct causal graph if interventions on all variables are available, despite not having any acyclicity constraints. Yet, we show in practice that the algorithm works on partial intervention sets as well. Fourthly, we extend ENCO to detecting latent confounders. In various experimental settings, ENCO recovers graphs accurately, making less than one error on graphs with 1,000 variables in less than nine hours of computation.
|
EFFICIENT NEURAL CAUSAL DISCOVERY WITHOUT ACYCLICITY CONSTRAINTS
|
d231648113
|
Neural Architecture Search (NAS) is quickly becoming the standard methodology to design neural network models. However, NAS is typically compute-intensive because multiple models need to be evaluated before choosing the best one. To reduce the computational power and time needed, a proxy task is often used for evaluating each model instead of full training. In this paper, we evaluate conventional reduced-training proxies and quantify how well they preserve ranking between multiple models during search when compared with the rankings produced by final trained accuracy. We propose a series of zero-cost proxies, based on recent pruning literature, that use just a single minibatch of training data to compute a model's score. Our zero-cost proxies use 3 orders of magnitude less computation but can match and even outperform conventional proxies. For example, Spearman's rank correlation coefficient between final validation accuracy and our best zero-cost proxy on NAS-Bench-201 is 0.82, compared to 0.61 for EcoNAS (a recently proposed reduced-training proxy). Finally, we use these zerocost proxies to enhance existing NAS search algorithms such as random search, reinforcement learning, evolutionary search and predictor-based search. For all search methodologies and across three different NAS datasets, we are able to significantly improve sample efficiency, and thereby decrease computation, by using our zero-cost proxies. For example on NAS-Bench-101, we achieved the same accuracy 4× quicker than the best previous result.
|
ZERO-COST PROXIES FOR LIGHTWEIGHT NAS
|
d47012356
|
As Machine Learning (ML) gets applied to security-critical or sensitive domains, there is a growing need for integrity and privacy for outsourced ML computations. A pragmatic solution comes from Trusted Execution Environments (TEEs), which use hardware and software protections to isolate sensitive computations from the untrusted software stack. However, these isolation guarantees come at a price in performance, compared to untrusted alternatives. This paper initiates the study of high performance execution of Deep Neural Networks (DNNs) in TEEs by efficiently partitioning DNN computations between trusted and untrusted devices. Building upon an efficient outsourcing scheme for matrix multiplication, we propose Slalom, a framework that securely delegates execution of all linear layers in a DNN from a TEE (e.g., Intel SGX or Sanctum) to a faster, yet untrusted, co-located processor. We evaluate Slalom by running DNNs in an Intel SGX enclave, which selectively delegates work to an untrusted GPU. For canonical DNNs (VGG16, MobileNet and ResNet variants) we obtain 6× to 20× increases in throughput for verifiable inference, and 4× to 11× for verifiable and private inference.Published as a conference paper at ICLR 2019This leads us to the main question of this paper: How can we most efficiently leverage TEEs for secure machine learning? This was posed byStoica et al. (2017)as one of nine open research problems for system challenges in AI. A specific challenge they raised is that of appropriately splitting ML computations between trusted and untrusted components, to increase efficiency as well as security by minimizing the Trusted Computing Base. This paper explores a novel approach to this challenge, wherein a Deep Neural Network (DNN) execution is partially outsourced from a TEE to a co-located, untrusted but faster device. Our approach, inspired by the verifiable ASICs of Wahby et al.(2016), differs from cryptographic ML outsourcing. In our case, work is delegated between two co-located parties, thus allowing for highly interactive-yet conceptually simpleroutsourcing protocols with orders-of-magnitude better efficiency. Our work also departs from prior systems that execute DNNs fully in a TEE(Ohrimenko et al., 2016;Hunt et al., 2018;Cheng et al., 2018;Hanzlik et al., 2018).The main observation that guides our approach is that matrix multiplication-the main bottleneck in DNNsadmits a concretely efficient verifiable outsourcing scheme known as Freivalds' algorithm(Freivalds, 1977), which can also be turned private in our setting. Our TEE selectively outsources these CPU intensive steps to a fast untrusted co-processor (and runs the remaining steps itself) therefore achieving much better performance than running the entire computation in the enclave, without compromising security.
|
SLALOM: FAST, VERIFIABLE AND PRIVATE EXECUTION OF NEURAL NETWORKS IN TRUSTED HARDWARE
|
d202573030
|
Nowadays, deep neural networks (DNNs) have become the main instrument for machine learning tasks within a wide range of domains, including vision, NLP, and speech. Meanwhile, in an important case of heterogenous tabular data, the advantage of DNNs over shallow counterparts remains questionable. In particular, there is no sufficient evidence that deep learning machinery allows constructing methods that outperform gradient boosting decision trees (GBDT), which are often the top choice for tabular problems. In this paper, we introduce Neural Oblivious Decision Ensembles (NODE), a new deep learning architecture, designed to work with any tabular data. In a nutshell, the proposed NODE architecture generalizes ensembles of oblivious decision trees, but benefits from both end-to-end gradient-based optimization and the power of multi-layer hierarchical representation learning. With an extensive experimental comparison to the leading GBDT packages on a large number of tabular datasets, we demonstrate the advantage of the proposed NODE architecture, which outperforms the competitors on most of the tasks. We open-source the PyTorch implementation of NODE and believe that it will become a universal framework for machine learning on tabular data.
|
NEURAL OBLIVIOUS DECISION ENSEMBLES FOR DEEP LEARNING ON TABULAR DATA
|
d56475997
|
The ability of a reinforcement learning (RL) agent to learn about many reward functions at the same time has many potential benefits, such as the decomposition of complex tasks into simpler ones, the exchange of information between tasks, and the reuse of skills. We focus on one aspect in particular, namely the ability to generalise to unseen tasks. Parametric generalisation relies on the interpolation power of a function approximator that is given the task description as input; one of its most common form are universal value function approximators (UVFAs). Another way to generalise to new tasks is to exploit structure in the RL problem itself. Generalised policy improvement (GPI) combines solutions of previous tasks into a policy for the unseen task; this relies on instantaneous policy evaluation of old policies under the new reward function, which is made possible through successor features (SFs). Our proposed universal successor features approximators (USFAs) combine the advantages of all of these, namely the scalability of UVFAs, the instant inference of SFs, and the strong generalisation of GPI. We discuss the challenges involved in training a USFA, its generalisation properties and demonstrate its practical benefits and transfer abilities on a large-scale domain in which the agent has to navigate in a first-person perspective three-dimensional environment.
|
UNIVERSAL SUCCESSOR FEATURES APPROXIMATORS
|
d227337121
|
Although spatio-temporal graph neural networks have achieved great empirical success in handling multiple correlated time series, they may be impractical in some real-world scenarios due to a lack of sufficient high-quality training data. Furthermore, spatio-temporal graph neural networks lack theoretical interpretation. To address these issues, we put forth a novel mathematically designed framework to analyze spatio-temporal data. Our proposed spatio-temporal graph scattering transform (ST-GST) extends traditional scattering transforms to the spatiotemporal domain. It performs iterative applications of spatio-temporal graph wavelets and nonlinear activation functions, which can be viewed as a forward pass of spatio-temporal graph convolutional networks without training. Since all the filter coefficients in ST-GST are mathematically designed, it is promising for the real-world scenarios with limited training data, and also allows for a theoretical analysis, which shows that the proposed ST-GST is stable to small perturbations of input signals and structures. Finally, our experiments show that i) ST-GST outperforms spatio-temporal graph convolutional networks by an increase of 35% in accuracy for MSR Action3D dataset; ii) it is better and computationally more efficient to design the transform based on separable spatio-temporal graphs than the joint ones; and iii) the nonlinearity in ST-GST is critical to empirical performance.
|
SPATIO-TEMPORAL GRAPH SCATTERING TRANSFORM
|
d13206339
|
Knowledge bases (KB), both automatically and manually constructed, are often incomplete -many valid facts can be inferred from the KB by synthesizing existing information. A popular approach to KB completion is to infer new relations by combinatory reasoning over the information found along other paths connecting a pair of entities. Given the enormous size of KBs and the exponential number of paths, previous path-based models have considered only the problem of predicting a missing relation given two entities, or evaluating the truth of a proposed triple. Additionally, these methods have traditionally used random paths between fixed entity pairs or more recently learned to pick paths between them. We propose a new algorithm, MINERVA 1 , which addresses the much more difficult and practical task of answering questions where the relation is known, but only one entity. Since random walks are impractical in a setting with combinatorially many destinations from a start node, we present a neural reinforcement learning approach which learns how to navigate the graph conditioned on the input query to find predictive paths. Empirically, this approach obtains state-of-the-art results on several datasets, significantly outperforming prior methods.
|
GO FOR A WALK AND ARRIVE AT THE ANSWER: REASONING OVER PATHS IN KNOWLEDGE BASES USING REINFORCEMENT LEARNING
|
d247058691
|
We introduce an adaptive tree search algorithm, that can find high-scoring outputs under translation models that make no assumptions about the form or structure of the search objective. This algorithm -a deterministic variant of Monte Carlo tree search -enables the exploration of new kinds of models that are unencumbered by constraints imposed to make decoding tractable, such as autoregressivity or conditional independence assumptions. When applied to autoregressive models, our algorithm has different biases than beam search has, which enables a new analysis of the role of decoding bias in autoregressive models. Empirically, we show that our adaptive tree search algorithm finds outputs with substantially better model scores compared to beam search in autoregressive models, and compared to reranking techniques in models whose scores do not decompose additively with respect to the words in the output. We also characterise the correlation of several translation model objectives with respect to BLEU. We find that while some standard models are poorly calibrated and benefit from the beam search bias, other often more robust models (autoregressive models tuned to maximize expected automatic metric scores, the noisy channel model and a newly proposed objective) benefit from increasing amounts of search using our proposed decoder, whereas the beam search bias limits the improvements obtained from such objectives. Thus, we argue that as models improve, the improvements may be masked by over-reliance on beam search or reranking based methods.
|
ENABLING ARBITRARY TRANSLATION OBJECTIVES WITH ADAPTIVE TREE SEARCH
|
d208527270
|
In Machine Learning as a Service, a provider trains a deep neural network and provides many users access to it. However, the hosted (source) model is susceptible to model stealing attacks where an adversary derives a surrogate model from API access to the source model. For post hoc detection of such attacks, the provider needs a robust method to determine whether a suspect model is a surrogate of their model or not. We propose a fingerprinting method for deep neural networks that extracts a set of inputs from the source model so that only surrogates agree with the source model on the classification of such inputs. These inputs are a specifically crafted subclass of targeted transferable adversarial examples which we call conferrable adversarial examples that transfer exclusively from a source model to its surrogates. We propose new methods to generate these conferrable adversarial examples and use them as our fingerprint. Our fingerprint is the first to be successfully tested as robust against distillation attacks, and our experiments show that this robustness extends to robustness against weaker removal attacks such as fine-tuning, ensemble attacks, and adversarial retraining. We even protect against a powerful adversary with white-box access to the source model, whereas the defender only needs black-box access to the surrogate model. We conduct our experiments on the CINIC dataset and a subset of ImageNet32 with 100 classes.
|
Deep Neural Network Fingerprinting by Conferrable Adversarial Examples
|
d233254411
|
Concept-based explanation approach is a popular model interpertability tool because it expresses the reasons for a model's predictions in terms of concepts that are meaningful for the domain experts. In this work, we study the problem of the concepts being correlated with confounding information in the features. We propose a new causal prior graph for modeling the impacts of unobserved variables and a method to remove the impact of confounding information and noise using a two-stage regression technique borrowed from the instrumental variable literature. We also model the completeness of the concepts set and show that our debiasing method works when the concepts are not complete. Our synthetic and real-world experiments demonstrate the success of our method in removing biases and improving the ranking of the concepts in terms of their contribution to the explanation of the predictions.
|
DEBIASING CONCEPT-BASED EXPLANATIONS WITH CAUSAL ANALYSIS
|
d3481593
|
Training deep neural networks requires many training samples, but in practice training labels are expensive to obtain and may be of varying quality, as some may be from trusted expert labelers while others might be from heuristics or other sources of weak supervision such as crowd-sourcing. This creates a fundamental qualityversus-quantity trade-off in the learning process. Do we learn from the small amount of high-quality data or the potentially large amount of weakly-labeled data? We argue that if the learner could somehow know and take the label-quality into account when learning the data representation, we could get the best of both worlds. To this end, we propose "fidelity-weighted learning" (FWL), a semi-supervised studentteacher approach for training deep neural networks using weakly-labeled data. FWL modulates the parameter updates to a student network (trained on the task we care about) on a per-sample basis according to the posterior confidence of its label-quality estimated by a teacher (who has access to the high-quality labels). Both student and teacher are learned from the data. We evaluate FWL on two tasks in information retrieval and natural language processing where we outperform state-of-the-art alternative semi-supervised methods, indicating that our approach makes better use of strong and weak labels, and leads to better task-dependent data representations.
|
FIDELITY-WEIGHTED LEARNING
|
d256627383
|
In this work, we attempt to bridge the two fields of finite-agent and infinite-agent games, by studying how the optimal policies of agents evolve with the number of agents (population size) in mean-field games, an agent-centric perspective in contrast to the existing works focusing typically on the convergence of the empirical distribution of the population. To this end, the premise is to obtain the optimal policies of a set of finite-agent games with different population sizes. However, either deriving the closed-form solution for each game is theoretically intractable, training a distinct policy for each game is computationally intensive, or directly applying the policy trained in a game to other games is sub-optimal. We address these challenges through the Population-size-Aware Policy Optimization (PAPO). Our contributions are three-fold. First, to efficiently generate efficient policies for games with different population sizes, we propose PAPO, which unifies two natural options (augmentation and hypernetwork) and achieves significantly better performance. PAPO consists of three components: i) the population-size encoding which transforms the original value of population size to an equivalent encoding to avoid training collapse, ii) a hypernetwork to generate a distinct policy for each game conditioned on the population size, and iii) the population size as an additional input to the generated policy. Next, we construct a multi-task-based training procedure to efficiently train the neural networks of PAPO by sampling data from multiple games with different population sizes. Finally, extensive experiments on multiple environments show the significant superiority of PAPO over baselines, and the analysis of the evolution of the generated policies further deepens our understanding of the two fields of finite-agent and infinite-agent games. . Ondemand high-capacity ride-sharing via dynamic trip-vehicle assignment. PNAS, 114(3):462-467, 2017.
|
POPULATION-SIZE-AWARE POLICY OPTIMIZATION FOR MEAN-FIELD GAMES
|
d257219926
|
Soft threshold pruning is among the cutting-edge pruning methods with state-ofthe-art performance 1 . However, previous methods either perform aimless searching on the threshold scheduler or simply set the threshold trainable, lacking theoretical explanation from a unified perspective. In this work, we reformulate soft threshold pruning as an implicit optimization problem solved using the Iterative Shrinkage-Thresholding Algorithm (ISTA), a classic method from the fields of sparse recovery and compressed sensing. Under this theoretical framework, all threshold tuning strategies proposed in previous studies of soft threshold pruning are concluded as different styles of tuning L 1 -regularization term. We further derive an optimal threshold scheduler through an in-depth study of threshold scheduling based on our framework. This scheduler keeps L 1 -regularization coefficient stable, implying a time-invariant objective function from the perspective of optimization. In principle, the derived pruning algorithm could sparsify any mathematical model trained via SGD. We conduct extensive experiments and verify its state-of-the-art performance on both Artificial Neural Networks (ResNet-50 and MobileNet-V1) and Spiking Neural Networks (SEW ResNet-18) on ImageNet datasets. On the basis of this framework, we derive a family of pruning methods, including sparsify-during-training, early pruning, and pruning at initialization. The code is available at https://github.com/Yanqi-Chen/LATS. . Sparse MRI: The application of compressed sensing for rapid MR imaging. Magnetic Resonance in Medicine, 58(6):1182-1195, 2007. Wolfgang Maass. Networks of spiking neurons: The third generation of neural network models. Neural Networks, 10(9):1659-1671, 1997. Bernard Martinet. Régularisation d'inéquations variationnelles par approximations successives. Revue Francaise d'informatique et de Recherche operationelle, 4:154-158, 1970. Risk, Rajit Manohar, and Dharmendra S. Modha. A million spiking-neuron integrated circuit with a scalable communication network and interface. Science, 345(6197):668-673, 2014. training of artificial neural networks with adaptive sparse connectivity inspired by network science. Nature communications, 9(1):1-12, 2018. . NP-hardness of 0 minimization problems: revision and extension to the non-negative setting. . Low-rank plus sparse matrix decomposition for accelerated dynamic MRI with separation of background and dynamic components. propagation: A sparsity inducing weight reparameterisation.
|
A UNIFIED FRAMEWORK FOR SOFT THRESHOLD PRUNING
|
d59316477
|
A fundamental question in reinforcement learning is whether model-free algorithms are sample efficient. Recently, Jin et al. [6] proposed a Q-learning algorithm with UCB exploration policy, and proved it has nearly optimal regret bound for finite-horizon episodic MDP. In this paper, we adapt Q-learning with UCB-exploration bonus to infinite-horizon MDP with discounted rewards without accessing a generative model. We show that the sample complexity of exploration of our algorithm is bounded byÕ( SA 2 (1−γ) 7 ). This improves the previously best known result of O( SA 4 (1−γ) 8 ) in this setting achieved by delayed Q-learning[13], and matches the lower bound in terms of as well as S and A except for logarithmic factors. * These two authors contributed equally †
|
Q-learning with UCB Exploration is Sample Efficient for Infinite-Horizon MDP
|
d202750230
|
Overparameterized transformer networks have obtained state of the art results in various natural language processing tasks, such as machine translation, language modeling, and question answering. These models contain hundreds of millions of parameters, necessitating a large amount of computation and making them prone to overfitting. In this work, we explore LayerDrop, a form of structured dropout, which has a regularization effect during training and allows for efficient pruning at inference time. In particular, we show that it is possible to select sub-networks of any depth from one large network without having to finetune them and with limited impact on performance. We demonstrate the effectiveness of our approach by improving the state of the art on machine translation, language modeling, summarization, question answering, and language understanding benchmarks. Moreover, we show that our approach leads to small BERT-like models of higher quality compared to training from scratch or using distillation.
|
REDUCING TRANSFORMER DEPTH ON DEMAND WITH STRUCTURED DROPOUT
|
d11428611
|
The concepts of unitary evolution matrices and associative memory have boosted the field of Recurrent Neural Networks (RNN) to state-of-the-art performance in a variety of sequential tasks. However, RNN still have a limited capacity to manipulate long-term memory. To bypass this weakness the most successful applications of RNN use external techniques such as attention mechanisms. In this paper we propose a novel RNN model that unifies the state-of-the-art approaches: Rotational Unit of Memory (RUM). The core of RUM is its rotational operation, which is, naturally, a unitary matrix, providing architectures with the power to learn long-term dependencies by overcoming the vanishing and exploding gradients problem. Moreover, the rotational unit also serves as associative memory. We evaluate our model on synthetic memorization, question answering and language modeling tasks. RUM learns the Copying Memory task completely and improves the state-of-the-art result in the Recall task. RUM's performance in the bAbI Question Answering task is comparable to that of models with attention mechanism. We also improve the state-of-the-art result to 1.189 bits-per-character (BPC) loss in the Character Level Penn Treebank (PTB) task, which is to signify the applications of RUM to real-world sequential data. The universality of our construction, at the core of RNN, establishes RUM as a promising approach to language modeling, speech recognition and machine translation. * equal contribution 1 arXiv:1710.09537v1 [cs.LG]
|
ROTATIONAL UNIT OF MEMORY
|
d203641721
|
A key element of understanding the efficacy of overparameterized neural networks is characterizing how they represent functions as the number of weights in the network approaches infinity. In this paper, we characterize the norm required to realize a function f : R d → R as a single hidden-layer ReLU network with an unbounded number of units (infinite width), but where the Euclidean norm of the weights is bounded, including precisely characterizing which functions can be realized with finite norm. This was settled for univariate functions f : R → R in Savarese et al.(2019), where it was shown that the required norm is determined by the L 1 -norm of the second derivative of the function. We extend the characterization to multi-variate functions (d ≥ 2, i.e., multiple input units), relating the required norm to the L 1 -norm of the Radon transform of a (d + 1)/2-power Laplacian of the function. This characterization allows us to show that all functions in Sobolev spaces W s,1 (R d ), s ≥ d + 1, can be represented with bounded norm, to calculate the required norm for several specific functions, and to obtain a depth separation result. These results have important implications for understanding generalization performance and the distinction between neural networks and more traditional kernel learning.
|
A FUNCTION SPACE VIEW OF BOUNDED NORM INFINITE WIDTH RELU NETS: THE MULTIVARIATE CASE
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.