_id stringlengths 4 10 | text stringlengths 0 18.4k | title stringlengths 0 8.56k |
|---|---|---|
d10296217 | This paper presents an actor-critic deep reinforcement learning agent with experience replay that is stable, sample efficient, and performs remarkably well on challenging environments, including the discrete 57-game Atari domain and several continuous control problems. To achieve this, the paper introduces several innovations, including truncated importance sampling with bias correction, stochastic dueling network architectures, and a new trust region policy optimization method.Published as a conference paper at ICLR 2017 is the first to address this challenge successfully at scale. More specifically, we introduce an actor critic with experience replay (ACER) that nearly matches the state-of-the-art performance of deep Q-networks with prioritized replay on Atari, and substantially outperforms A3C in terms of sample efficiency on both Atari and continuous control domains.ACER capitalizes on recent advances in deep neural networks, variance reduction techniques, the off-policy Retrace algorithm (Munos et al., 2016) and parallel training of RL agents (Mnih et al., 2016). Yet, crucially, its success hinges on innovations advanced in this paper: truncated importance sampling with bias correction, stochastic dueling network architectures, and efficient trust region policy optimization.On the theoretical front, the paper proves that the Retrace operator can be rewritten from our proposed truncated importance sampling with bias correction technique. | Published as a conference paper at ICLR 2017 SAMPLE EFFICIENT ACTOR-CRITIC WITH EXPERIENCE REPLAY |
d256846667 | Gradient clipping is an important technique for deep neural networks with exploding gradients, such as recurrent neural networks. Recent studies have shown that the loss functions of these networks do not satisfy the conventional smoothness condition, but instead satisfy a relaxed smoothness condition, i.e., the Lipschitz constant of the gradient scales linearly in terms of the gradient norm. Due to this observation, several gradient clipping algorithms have been developed for nonconvex and relaxed-smooth functions. However, the existing algorithms only apply to the single-machine or multiple-machine setting with homogeneous data across machines. It remains unclear how to design provably efficient gradient clipping algorithms in the general Federated Learning (FL) setting with heterogeneous data and limited communication rounds. In this paper, we design EPISODE, the very first algorithm to solve FL problems with heterogeneous data in the nonconvex and relaxed smoothness setting. The key ingredients of the algorithm are two new techniques called episodic gradient clipping and periodic resampled corrections. At the beginning of each round, EPISODE resamples stochastic gradients from each client and obtains the global averaged gradient, which is used to (1) determine whether to apply gradient clipping for the entire round and (2) construct local gradient corrections for each client. Notably, our algorithm and analysis provide a unified framework for both homogeneous and heterogeneous data under any noise level of the stochastic gradient, and it achieves state-of-the-art complexity results. In particular, we prove that EPISODE can achieve linear speedup in the number of machines, and it requires significantly fewer communication rounds. Experiments on several heterogeneous datasets, including text classification and image classification, show the superior performance of EPISODE over several strong baselines in FL. The code is available at https://github.com/MingruiLiu-ML-Lab/episode.Published as a conference paper at ICLR 2023 L0σ 2 (L0+L1(κ+σ))(1+ σ )N 2 Although there is a vast literature on FL (see (Kairouz et al., 2019) and references therein), the theoretical and algorithmic understanding of gradient clipping algorithms for training deep neural networks in the FL setting remains nascent. To the best of our knowledge, Liu et al. (2022) is the only work that has considered a communication-efficient distributed gradient clipping algorithm under the nonconvex and relaxed smoothness conditions in the FL setting. In particular, Liu et al. (2022) proved that their algorithm achieves linear speedup in terms of the number of clients and reduced communication rounds. Nevertheless, their algorithm and analysis are only applicable to the case of homogeneous data. In addition, the analyses of the stochastic gradient clipping algorithms in both single machine (Zhang et al., 2020a) and multiple-machine setting (Liu et al., 2022) require strong distributional assumptions on the stochastic gradient noise 3 , which may not hold in practice.In this work, we introduce a provably computation and communication efficient gradient clipping algorithm for nonconvex and relaxed-smooth functions in the general FL setting (i.e., heterogeneous data, limited communication) and without any distributional assumptions on the stochastic gradient noise. Compared with previous work on gradient clipping (Zhang et al., 2019a; 2020a; Cutkosky & Mehta, 2020; Liu et al., 2022) and FL with heterogeneous data (Li et al., 2020a; Karimireddy et al., 2020), our algorithm design relies on two novel techniques: episodic gradient clipping and periodic resampled corrections. In a nutshell, at the beginning of each communication round, the algorithm resamples each client's stochastic gradient; this information is used to decide whether to apply clipping in the current round (i.e., episodic gradient clipping), and to perform local corrections to each client's update (i.e., periodic resampled corrections). These techniques are very different compared with previous work on gradient clipping. Specifically, (1) In traditional gradient clipping (Pascanu et al., 2012; Zhang et al., 2019a; 2020a; Liu et al., 2022), whether or not to apply the clipping operation is determined only by the norm of the client's current stochastic gradient. Instead, we use the norm of the global objective's stochastic gradient (resampled at the beginning of the round) to determine whether or not clipping will be applied throughout the entire communication round. (2) Different from Karimireddy et al. (2020) which uses historical gradient information from the previous round to perform corrections, our algorithm utilizes the resampled gradient to correct each client's local update towards the global gradient, which mitigates the effect of data heterogeneity. Notice that, under the relaxed smoothness setting, the gradient may change quickly around a point at which the gradient norm is large. Therefore, our algorithm treats a small gradient as more "reliable" and confidently applies the unclipped corrected local updates; on the contrary, the algorithm regards a large gradient as less "reliable" and in this case clips the corrected local updates. Our major contributions are summarized as follows. Vien V Mai and Mikael Johansson. Stability and convergence of stochastic gradient clipping: Beyond lipschitz continuity and smoothness. arXiv preprint arXiv:2102.06489, 2021. H Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, et al. Communication-efficient learning of deep networks from decentralized data. AISTATS, 2017a. . Deep contextualized word representations. arXiv preprint arXiv:1802.05365, 2018. . Cooperative sgd: A unified framework for the design and analysis of communication-efficient sgd algorithms. arXiv preprint arXiv:1808.07576, 2018. Paul J Werbos. Generalization of backpropagation with application to a recurrent gas market model. . The min-max complexity of distributed stochastic convex optimization with intermittent communication. arXiv preprint arXiv:2102.01583, 2021. . On the linear speedup analysis of communication efficient momentum sgd for distributed non-convex optimization. arXiv preprint arXiv:1905.03817, 2019a. . On the linear speedup analysis of communication efficient momentum SGD for distributed non-convex optimization. In . Parallel restarted sgd with faster convergence and less communication: Demystifying why model averaging works for deep learning. In | EPISODE: EPISODIC GRADIENT CLIPPING WITH PE- RIODIC RESAMPLED CORRECTIONS FOR FEDERATED LEARNING WITH HETEROGENEOUS DATA |
d11994930 | We present a probabilistic variant of the recently introduced maxout unit. The success of deep neural networks utilizing maxout can partly be attributed to favorable performance under dropout, when compared to rectified linear units. It however also depends on the fact that each maxout unit performs a pooling operation over a group of linear transformations and is thus partially invariant to changes in its input. Starting from this observation we ask the question: Can the desirable properties of maxout units be preserved while improving their invariance properties ? We argue that our probabilistic maxout (probout) units successfully achieve this balance. We quantitatively verify this claim and report classification performance matching or exceeding the current state of the art on three challenging image classification benchmarks (CIFAR-10, CIFAR-100 and SVHN). | Improving Deep Neural Networks with Probabilistic Maxout Units |
d231933751 | Performing inference on deep learning models for videos remains a challenge due to the large amount of computational resources required to achieve robust recognition. An inherent property of real-world videos is the high correlation of information across frames which can translate into redundancy in either temporal or spatial feature maps of the models, or both. The type of redundant features depends on the dynamics and type of events in the video: static videos have more temporal redundancy while videos focusing on objects tend to have more channel redundancy. Here we present a redundancy reduction framework, termed VA-RED 2 , which is input-dependent. Specifically, our VA-RED 2 framework uses an input-dependent policy to decide how many features need to be computed for temporal and channel dimensions. To keep the capacity of the original model, after fully computing the necessary features, we reconstruct the remaining redundant features from those using cheap linear operations. We learn the adaptive policy jointly with the network weights in a differentiable way with a shared-weight mechanism, making it highly efficient. Extensive experiments on multiple video datasets and different visual tasks show that our framework achieves 20% − 40% reduction in computation (FLOPs) when compared to state-of-the-art methods without any performance loss. | VA-RED 2 : VIDEO ADAPTIVE REDUNDANCY REDUC- TION |
d211132756 | While recent continual learning methods largely alleviate the catastrophic problem on toy-sized datasets, some issues remain to be tackled to apply them to real-world problem domains. First, a continual learning model should effectively handle catastrophic forgetting and be efficient to train even with a large number of tasks. Secondly, it needs to tackle the problem of order-sensitivity, where the performance of the tasks largely varies based on the order of the task arrival sequence, as it may cause serious problems where fairness plays a critical role (e.g. medical diagnosis). To tackle these practical challenges, we propose a novel continual learning method that is scalable as well as order-robust, which instead of learning a completely shared set of weights, represents the parameters for each task as a sum of task-shared and sparse task-adaptive parameters. With our Additive Parameter Decomposition (APD), the task-adaptive parameters for earlier tasks remain mostly unaffected, where we update them only to reflect the changes made to the task-shared parameters. This decomposition of parameters effectively prevents catastrophic forgetting and order-sensitivity, while being computation-and memory-efficient. Further, we can achieve even better scalability with APD using hierarchical knowledge consolidation, which clusters the task-adaptive parameters to obtain hierarchically shared parameters. We validate our network with APD, APD-Net, on multiple benchmark datasets against state-of-the-art continual learning methods, which it largely outperforms in accuracy, scalability, and order-robustness. | Published as a conference paper at ICLR 2020 SCALABLE AND ORDER-ROBUST CONTINUAL LEARN- ING WITH ADDITIVE PARAMETER DECOMPOSITION |
d3280771 | The driving force behind deep networks is their ability to compactly represent rich classes of functions. The primary notion for formally reasoning about this phenomenon is expressive efficiency, which refers to a situation where one network must grow unfeasibly large in order to realize (or approximate) functions of another. To date, expressive efficiency analyses focused on the architectural feature of depth, showing that deep networks are representationally superior to shallow ones. In this paper we study the expressive efficiency brought forth by connectivity, motivated by the observation that modern networks interconnect their layers in elaborate ways. We focus on dilated convolutional networks, a family of deep models delivering state of the art performance in sequence processing tasks. By introducing and analyzing the concept of mixed tensor decompositions, we prove that interconnecting dilated convolutional networks can lead to expressive efficiency. In particular, we show that even a single connection between intermediate layers can already lead to an almost quadratic gap, which in large-scale settings typically makes the difference between a model that is practical and one that is not. Empirical evaluation demonstrates how the expressive efficiency of connectivity, similarly to that of depth, translates into gains in accuracy. This leads us to believe that expressive efficiency may serve a key role in the development of new tools for deep network design. COHEN TAMARI SHASHUA Shamir (2015); Cohen et al. (2016b); Cohen and Shashua (2016a); Poggio et al. (2015); Mhaskar et al. (2016)) have focused on the architectural feature of depth, showing instances where deep networks are expressively efficient w.r.t. shallow ones. This theoretical focus is motivated by the vast empirical evidence supporting the importance of depth (see LeCun et al. (2015) for a survey of such results). However, it largely overlooks an additional architectural feature that in recent years is proving to have great impact on the performance of deep networks -connectivity. Nearly all state of the art networks these days (e.g. Szegedy et al. (2015); He et al. (2015); Huang et al. (2016b,a)) deviate from the simple feed-forward (chain) approach, running layers connected under various schemes. Whether or not this relates to expressive efficiency remains to be an open question. A specific family of deep networks gaining increased attention in the deep learning community is that of dilated convolutional networks. These models form the basis of the recent WaveNet (van den Oord et al. (2016)) and ByteNet (Kalchbrenner et al. (2016)) architectures, which provide state of the art performance in audio and text processing tasks. Dilated convolutional networks are typically applied to sequence data, and consist of multiple succeeding convolutional layers, each comprising non-contiguous filters with a different dilation (distance between neighboring elements). The choice of dilations directly affects the space of functions that may be realized by a network, and while no choice is expressively efficient w.r.t. another, we show in this work that interconnecting networks with different dilations leads to expressive efficiency, and by this demonstrate that connectivity indeed bears the potential to enhance the expressiveness of deep networks.Our analysis follows several recent works utilizing tensor decompositions for theoretical studies of deep learning (see for exampleJanzamin et al. (2015); Sedghi and Anandkumar (2016)), and in particular, builds on the equivalence between hierarchical tensor decompositions and convolutional networks established in and . We show that with dilated convolutional networks, the choice of dilations throughout a network corresponds to determination of the mode (dimension) tree underlying the respective decomposition. We then define the notion of a mixed tensor decomposition, which blends together multiple mode trees, effectively creating a large ensemble of hybrid trees formed from all possible combinations. Mixed tensor decompositions correspond to mixed dilated convolutional networks, i.e. mixtures formed by connecting intermediate layers of different dilated convolutional networks. This allows studying the expressive properties of such mixtures using mathematical machinery from the field of tensor analysis. We fully analyze a particular case of dilated convolutional arithmetic circuits, showing that a single connection between intermediate layers already leads to an almost quadratic expressive efficiency, which in large-scale settings typically makes the difference between a model that is practical and one that is not.An experiment on TIMIT speech recognition dataset(Garofolo et al. (1993)) evaluates the dilated convolutional network architectures covered by our analysis. We find that adding connections between intermediate layers of different networks improves accuracy, with no additional cost in terms of computation or model capacity. This serves as an indication that with the architectural feature of connectivity, similarly to the case of depth, expressive efficiency and improved accuracies go hand in hand. Accordingly, we believe expressive efficiency may serve a key role in the development of new tools for deep network design.Summary of Our Analysis and ContributionsOur analysis begins in sec. 4, where we present the dilated convolutional network underlying WaveNet (fig. 1). We consider this to be the baseline architecture and, following Cohen and Shashua | Boosting Dilated Convolutional Networks with Mixed Tensor Decompositions Amnon Shashua |
d252693021 | We consider infinite-horizon discounted Markov decision processes and study the convergence rates of the natural policy gradient (NPG) and the Q-NPG methods with the log-linear policy class. Using the compatible function approximation framework, both methods with log-linear policies can be written as inexact versions of the policy mirror descent (PMD) method. We show that both methods attain linear convergence rates andÕ(1/ǫ 2 ) sample complexities using a simple, non-adaptive geometrically increasing step size, without resorting to entropy or other strongly convex regularization. Lastly, as a byproduct, we obtain sublinear convergence rates for both methods with arbitrary constant step size.keywords discounted Markov decision process, natural policy gradient, policy mirror descent, log-linear policy, sample complexity. | Linear Convergence of Natural Policy Gradient Methods with Log-Linear Policies * |
d214184365 | Differentiable architecture search (DARTS) provided a fast solution in finding effective network architectures, but suffered from large memory and computing overheads in jointly training a super-network and searching for an optimal architecture. In this paper, we present a novel approach, namely, Partially-Connected DARTS, by sampling a small part of super-network to reduce the redundancy in exploring the network space, thereby performing a more efficient search without comprising the performance. In particular, we perform operation search in a subset of channels while bypassing the held out part in a shortcut. This strategy may suffer from an undesired inconsistency on selecting the edges of super-net caused by sampling different channels. We alleviate it using edge normalization, which adds a new set of edge-level parameters to reduce uncertainty in search. Thanks to the reduced memory cost, PC-DARTS can be trained with a larger batch size and, consequently, enjoys both faster speed and higher training stability. Experimental results demonstrate the effectiveness of the proposed method. Specifically, we achieve an error rate of 2.57% on CIFAR10 with merely 0.1 GPU-days for architecture search, and a state-of-the-art top-1 error rate of 24.2% on ImageNet (under the mobile setting) using 3.8 GPU-days for search. Our code has been made available at https://github.com/yuhuixu1993/PC-DARTS. | Published as a conference paper at ICLR 2020 PC-DARTS: PARTIAL CHANNEL CONNECTIONS FOR MEMORY-EFFICIENT ARCHITECTURE SEARCH |
d6954272 | Sequence to sequence learning has recently emerged as a new paradigm in supervised learning. To date, most of its applications focused on only one task and not much work explored this framework for multiple tasks. This paper examines three multi-task learning (MTL) settings for sequence to sequence models: (a) the oneto-many setting -where the encoder is shared between several tasks such as machine translation and syntactic parsing, (b) the many-to-one setting -useful when only the decoder can be shared, as in the case of translation and image caption generation, and (c) the many-to-many setting -where multiple encoders and decoders are shared, which is the case with unsupervised objectives and translation. Our results show that training on a small amount of parsing and image caption data can improve the translation quality between English and German by up to 1.5 BLEU points over strong single-task baselines on the WMT benchmarks. Furthermore, we have established a new state-of-the-art result in constituent parsing with 93.0 F 1 . Lastly, we reveal interesting properties of the two unsupervised learning objectives, autoencoder and skip-thought, in the MTL context: autoencoder helps less in terms of perplexities but more on BLEU scores compared to skip-thought. * Minh-Thang Luong is also a student at Stanford University. | MULTI-TASK SEQUENCE TO SEQUENCE LEARNING |
d67788344 | Reinforcement learning is a promising framework for solving control problems, but its use in practical situations is hampered by the fact that reward functions are often difficult to engineer. Specifying goals and tasks for autonomous machines, such as robots, is a significant challenge: conventionally, reward functions and goal states have been used to communicate objectives. But people can communicate objectives to each other simply by describing or demonstrating them. How can we build learning algorithms that will allow us to tell machines what we want them to do? In this work, we investigate the problem of grounding language commands as reward functions using inverse reinforcement learning, and argue that language-conditioned rewards are more transferable than language-conditioned policies to new environments. We propose language-conditioned reward learning (LC-RL), which grounds language commands as a reward function represented by a deep neural network. We demonstrate that our model learns rewards that transfer to novel tasks and environments on realistic, high-dimensional visual environments with natural language commands, whereas directly learning a languageconditioned policy leads to poor performance. * Work done during an internship at Google AI Research | FROM LANGUAGE TO GOALS: INVERSE REINFORCE- MENT LEARNING FOR VISION-BASED INSTRUCTION FOLLOWING |
d3604396 | Neural networks are known to be vulnerable to adversarial examples. Carefully chosen perturbations to real images, while imperceptible to humans, induce misclassification and threaten the reliability of deep learning systems in the wild. To guard against adversarial examples, we take inspiration from game theory and cast the problem as a minimax zero-sum game between the adversary and the model. In general, for such games, the optimal strategy for both players requires a stochastic policy, also known as a mixed strategy. In this light, we propose Stochastic Activation Pruning (SAP), a mixed strategy for adversarial defense. SAP prunes a random subset of activations (preferentially pruning those with smaller magnitude) and scales up the survivors to compensate. We can apply SAP to pretrained networks, including adversarially trained models, without fine-tuning, providing robustness against adversarial examples. Experiments demonstrate that SAP confers robustness against attacks, increasing accuracy and preserving calibration. | Published as a conference paper at ICLR 2018 STOCHASTIC ACTIVATION PRUNING FOR ROBUST ADVERSARIAL DEFENSE |
d246411466 | We present a novel neural-networks-based algorithm to compute optimal transport maps and plans for strong and weak transport costs. To justify the usage of neural networks, we prove that they are universal approximators of transport plans between probability distributions. We evaluate the performance of our optimal transport algorithm on toy examples and on the unpaired image-to-image translation.(a) Celeba (female) → anime, outdoor → church, deterministic (one-to-one, W2).(b) Handbags → shoes, stochastic (one-to-many, W2,1).Published as a conference paper at ICLR 2023 not exist. (Daniels et al., 2021) recover the entropy-regularized stochastic plan, but the procedures for learning the plan and sampling from it are extremely time-consuming due to using score-based models and the Langevin dynamics (Daniels et al., 2021, 6). Contributions. We propose a novel algorithm to compute deterministic and stochastic OT plans with deep neural networks ( 4.1, 4.2). Our algorithm is designed for weak and strong optimal transport costs ( 2) and generalizes previously known scalable approaches ( 3, 4.3). To reinforce the usage of neural nets, we prove that they are universal approximators of transport plans ( 4.4). We show that our algorithm can be applied to large-scale computer vision tasks ( 5).Notations. We use X , Y, Z to denote Polish spaces and P(X ), P(Y), P(Z) to denote the respective sets of probability distributions on them. We denote the set of probability distributions on X × Y with marginals P and Q by Π(P, Q). For a measurable map T : X × Z → Y (or T : X → Y), we denote the associated push-forward operator by T # . | Published as a conference paper at ICLR 2023 NEURAL OPTIMAL TRANSPORT |
d15310708 | The bag-of-words (BOW) model is the common approach for classifying documents, where words are used as feature for training a classifier. This generally involves a huge number of features. Some techniques, such as Latent Semantic Analysis (LSA) or Latent Dirichlet Allocation (LDA), have been designed to summarize documents in a lower dimension with the least semantic information loss. Some semantic information is nevertheless always lost, since only words are considered. Instead, we aim at using information coming from n-grams to overcome this limitation, while remaining in a low-dimension space. Many approaches, such as the Skip-gram model, provide good word vector representations very quickly. We propose to average these representations to obtain representations of n-grams. All n-grams are thus embedded in a same semantic space. A K-means clustering can then group them into semantic concepts. The number of features is therefore dramatically reduced and documents can be represented as bag of semantic concepts. We show that this model outperforms LSA and LDA on a sentiment classification task, and yields similar results than a traditional BOW-model with far less features. | Under review as a conference paper at ICLR 2015 N-GRAM-BASED LOW-DIMENSIONAL REPRESENTA- TION FOR DOCUMENT CLASSIFICATION |
d235265790 | Clustering is one of the most fundamental tasks in machine learning. Recently, deep clustering has become a major trend in clustering techniques. Representation learning often plays an important role in the effectiveness of deep clustering, and thus can be a principal cause of performance degradation. In this paper, we propose a clustering-friendly representation learning method using instance discrimination and feature decorrelation. Our deep-learning-based representation learning method is motivated by the properties of classical spectral clustering. Instance discrimination learns similarities among data and feature decorrelation removes redundant correlation among features. We utilize an instance discrimination method in which learning individual instance classes leads to learning similarity among instances. Through detailed experiments and examination, we show that the approach can be adapted to learning a latent space for clustering. We design novel softmax-formulated decorrelation constraints for learning. In evaluations of image clustering using CIFAR-10 and ImageNet-10, our method achieves accuracy of 81.5% and 95.4%, respectively. We also show that the softmax-formulated constraints are compatible with various neural networks. | CLUSTERING-FRIENDLY REPRESENTATION LEARN- ING VIA INSTANCE DISCRIMINATION AND FEATURE DECORRELATION |
d18882526 | This paper presents experiments extending the work ofBa et al. (2014)on recurrent neural models for attention into less constrained visual environments, beginning with fine-grained categorization on the Stanford Dogs data set. In this work we use an RNN of the same structure but substitute a more powerful visual network and perform large-scale pre-training of the visual network outside of the attention RNN. Most work in attention models to date focuses on tasks with toy or more constrained visual environments. We present competitive results for finegrained categorization. More importantly, we show that our model learns to direct high resolution attention to the most discriminative regions without any spatial supervision such as bounding boxes. Given a small input window, it is hence able to discriminate fine-grained dog breeds with cheap glances at faces and fur patterns, while avoiding expensive and distracting processing of entire images. In addition to allowing high resolution processing with a fixed budget and naturally handling static or sequential inputs, this approach has the major advantage of being trained end-to-end, unlike most current approaches which are heavily engineered. | Under review as a workshop contribution at ICLR 2015 ATTENTION FOR FINE-GRAINED CATEGORIZATION |
d9062199 | We introduce a two-layer wavelet scattering network, for object classification. This scattering transform computes a spatial wavelet transform on the first layer and a new joint wavelet transform along spatial, angular and scale variables in the second layer. Numerical experiments demonstrate that this two layer convolution network, which involves no learning and no max pooling, performs efficiently on complex image data sets such as CalTech, with structural objects variability and clutter. It opens the possibility to simplify deep neural network learning by initializing the first layers with wavelet filters. | Generic Deep Networks with Wavelet Scattering |
d258436929 | Simulating the time evolution of physical systems is pivotal in many scientific and engineering problems. An open challenge in simulating such systems is their multi-resolution dynamics: a small fraction of the system is extremely dynamic, and requires very fine-grained resolution, while a majority of the system is changing slowly and can be modeled by coarser spatial scales. Typical learning-based surrogate models use a uniform spatial scale, which needs to resolve to the finest required scale and can waste a huge compute to achieve required accuracy. In this work, we introduce Learning controllable Adaptive simulation for Multiresolution Physics (LAMP) as the first full deep learning-based surrogate model that jointly learns the evolution model and optimizes appropriate spatial resolutions that devote more compute to the highly dynamic regions. LAMP consists of a Graph Neural Network (GNN) for learning the forward evolution, and a GNNbased actor-critic for learning the policy of spatial refinement and coarsening. We introduce learning techniques that optimizes LAMP with weighted sum of error and computational cost as objective, allowing LAMP to adapt to varying relative importance of error vs. computation tradeoff at inference time. We evaluate our method in a 1D benchmark of nonlinear PDEs and a challenging 2D mesh-based simulation. We demonstrate that our LAMP outperforms state-of-the-art deep learning surrogate models, and can adaptively trade-off computation to improve long-term prediction error: it achieves an average of 33.7% error reduction for 1D nonlinear PDEs, and outperforms MeshGraphNets + classical Adaptive Mesh Refinement (AMR) in 2D mesh-based simulations. Project website with data and code can be found at: | Published as a conference paper at ICLR 2023 LEARNING CONTROLLABLE ADAPTIVE SIMULATION FOR MULTI-RESOLUTION PHYSICS |
d255998264 | Classical wisdom suggests that estimators should avoid fitting noise to achieve good generalization. In contrast, modern overparameterized models can yield small test error despite interpolating noise -a phenomenon often called "benign overfitting" or "harmless interpolation". This paper argues that the degree to which interpolation is harmless hinges upon the strength of an estimator's inductive bias, i.e., how heavily the estimator favors solutions with a certain structure: while strong inductive biases prevent harmless interpolation, weak inductive biases can even require fitting noise to generalize well. Our main theoretical result establishes tight non-asymptotic bounds for high-dimensional kernel regression that reflect this phenomenon for convolutional kernels, where the filter size regulates the strength of the inductive bias. We further provide empirical evidence of the same behavior for deep neural networks with varying filter sizes and rotational invariance. * Equal contribution; correspondence to research@michaelaerni.com | Published as a conference paper at ICLR 2023 STRONG INDUCTIVE BIASES PROVABLY PREVENT HARMLESS INTERPOLATION |
d14482228 | This paper introduces a greedy parser based on neural networks, which leverages a new compositional sub-tree representation. The greedy parser and the compositional procedure are jointly trained, and tightly depends on each-other. The composition procedure outputs a vector representation which summarizes syntactically (parsing tags) and semantically (words) sub-trees. Composition and tagging is achieved over continuous (word or tag) representations, and recurrent neural networks. We reach F1 performance on par with well-known existing parsers, while having the advantage of speed, thanks to the greedy nature of the parser. A fully functional parser implementing the method described on this paper can be obtained from people.idiap.ch/jlegrand/download. | JOINT RNN-BASED GREEDY PARSING AND WORD COMPOSITION |
d222378172 | Modern deep learning models have achieved great success in predictive accuracy for many data modalities. However, their application to many real-world tasks is restricted by poor uncertainty estimates, such as overconfidence on out-ofdistribution (OOD) data and ungraceful failing under distributional shift. Previous benchmarks have found that ensembles of neural networks (NNs) are typically the best calibrated models on OOD data. Inspired by this, we leverage recent theoretical advances that characterize the function-space prior of an ensemble of infinitelywide NNs as a Gaussian process, termed the neural network Gaussian process (NNGP). We use the NNGP with a softmax link function to build a probabilistic model for multi-class classification and marginalize over the latent Gaussian outputs to sample from the posterior. This gives us a better understanding of the implicit prior NNs place on function space and allows a direct comparison of the calibration of the NNGP and its finite-width analogue. We also examine the calibration of previous approaches to classification with the NNGP, which treat classification problems as regression to the one-hot labels. In this case the Bayesian posterior is exact, and we compare several heuristics to generate a categorical distribution over classes. We find these methods are well calibrated under distributional shift. Finally, we consider an infinite-width final layer in conjunction with a pre-trained embedding. This replicates the important practical use case of transfer learning and allows scaling to significantly larger datasets. As well as achieving competitive predictive accuracy, this approach is better calibrated than its finite width analogue. * Authors contributed equally to this work. † Work done as a member of the Google AI Residency program (https://g.co/airesidency). | EXPLORING THE UNCERTAINTY PROPERTIES OF NEU- RAL NETWORKS' IMPLICIT PRIORS IN THE INFINITE- WIDTH LIMIT |
d225040612 | Many problems across computer vision and the natural sciences require the analysis of spherical data, for which representations may be learned efficiently by encoding equivariance to rotational symmetries. We present a generalized spherical CNN framework that encompasses various existing approaches and allows them to be leveraged alongside each other. The only existing non-linear spherical CNN layer that is strictly equivariant has complexity OpC 2 L 5 q, where C is a measure of representational capacity and L the spherical harmonic bandlimit. Such a high computational cost often prohibits the use of strictly equivariant spherical CNNs. We develop two new strictly equivariant layers with reduced complexity OpCL 4 q and OpCL 3 log Lq, making larger, more expressive models computationally feasible. Moreover, we adopt efficient sampling theory to achieve further computational savings. We show that these developments allow the construction of more expressive hybrid models that achieve state-of-the-art accuracy and parameter efficiency on spherical benchmark problems. | Published as a conference paper at ICLR 2021 EFFICIENT GENERALIZED SPHERICAL CNNS |
d13468104 | Current work in lexical distributed representations maps each word to a point vector in low-dimensional space. Mapping instead to a density provides many interesting advantages, including better capturing uncertainty about a representation and its relationships, expressing asymmetries more naturally than dot product or cosine similarity, and enabling more expressive parameterization of decision boundaries. This paper advocates for density-based distributed embeddings and presents a method for learning representations in the space of Gaussian distributions. We compare performance on various word embedding benchmarks, investigate the ability of these embeddings to model entailment and other asymmetric relationships, and explore novel properties of the representation. | WORD REPRESENTATIONS VIA GAUSSIAN EMBEDDING |
d246823211 | We argue that a form of the valuable information provided by the auxiliary information is its implied data clustering information. For instance, considering hashtags as auxiliary information, we can hypothesize that an Instagram image will be semantically more similar with the same hashtags. With this intuition, we present a two-stage weakly-supervised contrastive learning approach. The first stage is to cluster data according to its auxiliary information. The second stage is to learn similar representations within the same cluster and dissimilar representations for data from different clusters. Our empirical experiments suggest the following three contributions. First, compared to conventional self-supervised representations, the auxiliary-information-infused representations bring the performance closer to the supervised representations, which use direct downstream labels as supervision signals. Second, our approach performs the best in most cases, when comparing our approach with other baseline representation learning methods that also leverage auxiliary data information. Third, we show that our approach also works well with unsupervised constructed clusters (e.g., no auxiliary information), resulting in a strong unsupervised representation learning approach. † Equal contribution. Code available at: https://github.com/Crazy-Jack/Cl-InfoNCE. | LEARNING WEAKLY-SUPERVISED CONTRASTIVE REP- RESENTATIONS |
d256901166 | Graph neural network (GNN) is a powerful learning approach for graph-based recommender systems. Recently, GNNs integrated with contrastive learning have shown superior performance in recommendation with their data augmentation schemes, aiming at dealing with highly sparse data. Despite their success, most existing graph contrastive learning methods either perform stochastic augmentation (e.g., node/edge perturbation) on the user-item interaction graph, or rely on the heuristic-based augmentation techniques (e.g., user clustering) for generating contrastive views. We argue that these methods cannot well preserve the intrinsic semantic structures and are easily biased by the noise perturbation. In this paper, we propose a simple yet effective graph contrastive learning paradigm LightGCL that mitigates these issues impairing the generality and robustness of CL-based recommenders. Our model exclusively utilizes singular value decomposition for contrastive augmentation, which enables the unconstrained structural refinement with global collaborative relation modeling. Experiments conducted on several benchmark datasets demonstrate the significant improvement in performance of our model over the state-of-the-arts. Further analyses demonstrate the superiority of LightGCL's robustness against data sparsity and popularity bias. The source code of our model is available at https://github.com/HKUDS/LightGCL. | LIGHTGCL: SIMPLE YET EFFECTIVE GRAPH CON- TRASTIVE LEARNING FOR RECOMMENDATION |
d238743879 | We give simpler, sparser, and faster algorithms for differentially private fine-tuning of large-scale pre-trained language models, which achieve the state-of-the-art privacy versus utility tradeoffs on many standard NLP tasks. We propose a meta-framework for this problem, inspired by the recent success of highly parameter-efficient methods for fine-tuning. Our experiments show that differentially private adaptations of these approaches outperform previous private algorithms in three important dimensions: utility, privacy, and the computational and memory cost of private training. On many commonly studied datasets, the utility of private models approaches that of non-private models. For example, on the MNLI dataset we achieve an accuracy of 87.8% using RoBERTa-Large and 83.5% using RoBERTa-Base with a privacy budget of ε = 6.7. In comparison, absent privacy constraints, RoBERTa-Large achieves an accuracy of 90.2%. Our findings are similar for natural language generation tasks. Privately fine-tuning with DART, GPT-2-Small, GPT-2-Medium, GPT-2-Large, and GPT-2-XL achieve BLEU scores of 38.5, 42.0, 43.1, and 43.8 respectively (privacy budget of ε = 6.8, δ = 1e-5) whereas the non-private baseline is 48.1. All our experiments suggest that larger models are better suited for private fine-tuning: while they are well known to achieve superior accuracy non-privately, we find that they also better maintain their accuracy when privacy is introduced. * Aside from the first and second authors, all other authors are listed in alphabetical order. † Sun Yat-sen University. yuda3@mail2.sysu.edu.cn. Work was done while an intern at Microsoft Research Asia. ‡ Microsoft. | Differentially Private Fine-tuning of Language Models * |
d7454685 | 1arXiv:1511.03962v4[cs.CL]
Text documents are structured on multiple levels of detail: individual words are related by syntax, and larger units of text are related by discourse structure.Existing language models generally fail to account for discourse structure, but it is crucial if we are to have language models that reward coherence and generate coherent texts.We present and empirically evaluate a set of multi-level recurrent neural network language models, called Document-Context Language Models (DCLMs), which incorporate contextual information both within and beyond the sentence.In comparison with sentence-level recurrent neural network language models, the DCLMs obtain slightly better predictive likelihoods, and considerably better assessment | DOCUMENT CONTEXT LANGUAGE MODELS
21 Feb 2016
Yangfeng Ji |
d227240564 | The largely successful method of training neural networks is to learn their weights using some variant of stochastic gradient descent (SGD). Here, we show that the solutions found by SGD can be further improved by ensembling a subset of the weights in late stages of learning. At the end of learning, we obtain back a single model by taking a spatial average in weight space. To avoid incurring increased computational costs, we investigate a family of low-dimensional late-phase weight models which interact multiplicatively with the remaining parameters. Our results show that augmenting standard models with late-phase weights improves generalization in established benchmarks such as CIFAR-10/100, ImageNet and enwik8. These findings are complemented with a theoretical analysis of a noisy quadratic problem which provides a simplified picture of the late phases of neural network learning. | Published as a conference paper at ICLR 2021 NEURAL NETWORKS WITH LATE-PHASE WEIGHTS |
d15461305 | Kernel density estimation, a.k.a. Parzen windows, is a popular density estimation method, which can be used for outlier detection or clustering. With multivariate data, its performance is heavily reliant on the metric used within the kernel. Most earlier work has focused on learning only the bandwidth of the kernel (i.e., a scalar multiplicative factor). In this paper, we propose to learn a full Euclidean metric through an expectation-minimisation (EM) procedure, which can be seen as an unsupervised counterpart to neighbourhood component analysis (NCA). In order to avoid overfitting with a fully nonparametric density estimator in high dimensions, we also consider a semi-parametric Gaussian-Parzen density model, where some of the variables are modelled through a jointly Gaussian density, while others are modelled through Parzen windows. For these two models, EM leads to simple closed-form updates based on matrix inversions and eigenvalue decompositions. We show empirically that our method leads to density estimators with higher test-likelihoods than natural competing methods, and that the metrics may be used within most unsupervised learning techniques that rely on local distances, such as spectral clustering or manifold learning methods. Finally, we present a stochastic approximation scheme which allows for the use of this method in a large-scale setting. | Local Component Analysis |
d245704549 | In model-free deep reinforcement learning (RL) algorithms, using noisy value estimates to supervise policy evaluation and optimization is detrimental to the sample efficiency. As this noise is heteroscedastic, its effects can be mitigated using uncertainty-based weights in the optimization process. Previous methods rely on sampled ensembles, which do not capture all aspects of uncertainty. We provide a systematic analysis of the sources of uncertainty in the noisy supervision that occurs in RL, and introduce inverse-variance RL, a Bayesian framework which combines probabilistic ensembles and Batch Inverse Variance weighting. We propose a method whereby two complementary uncertainty estimation methods account for both the Q-value and the environment stochasticity to better mitigate the negative impacts of noisy supervision. Our results show significant improvement in terms of sample efficiency on discrete and continuous control tasks. * Canada CIFAR AI Chair arXiv:2201.01666v3 [cs.LG] 3 May 2022Published as a conference paper at ICLR 2022 Distributional RL (Bellemare et al., 2017) considers the value function as a distribution to be learned as such. It is orthogonal to our proposition: we consider the uncertainty of the labels used to learn a scalar value function. In the offline RL setting, where the dataset is limited, uncertainty-weighted actor-critic (UWAC) (Wu et al., 2021) uses inverse-variance weighting to discard out-of-distribution state-action pairs using Monte Carlo dropout (Gal & Ghahramani, 2016) for uncertainty estimation.Closer to our work, propose SUNRISE, in which each sample of the Bellman backup in the TD update step is weighted to lower the importance of the targets which have a high standard deviation. The weights w(s , a ) are computed based on a sigmoïd of the negative standard deviationQ std (s , a ) scaled by a temperature hyperparameter T , and then offset such that they are between 0.5 and 1: w(s, a) = σ(−Q std (s , a ) * T ) + 0.5. The uncertainty of the target is estimated by sampled ensembles. While SUNRISE proposes other contributions such as an exploration bonus, the heuristic weighting scheme and the limitations of sampled ensembles in capturing the predictive uncertainty leave space for improvement in the mitigation of the effects of unreliable supervision.We propose inverse-variance reinforcement learning (IV-RL). IV-RL also uses weights to reduce the importance of uncertain targets in training. It does so by addressing the problem from two viewpoints. First, we use variance networks (Kendall & Gal, 2017), whose loss function for regression is the negative log-likelihood instead of the L2 distance. For a given state-action pair (s, a), the network learns the target's noise, due for example to the stochasticity of the environment or the update of the policy. It then naturally down-weights the highly noisy samples in the training process. Second, we use variance ensembles(Lakshminarayanan et al., 2017)to estimate the uncertainty of the target due to the prediction of Q(s , a ) during the temporal-difference update. We merge the predicted variances of several variance networks through a mixture of Gaussians, which has been shown to be a reliable method to capture predictive uncertainty (Ovadia et al., 2019). We then use Batch Inverse-Variance (BIV)(Mai et al., 2021), which has been shown to significantly improve the performance of supervised learning with neural networks in the case of heteroscedastic regression. BIV is normalized, which makes it ideal to cope with different and time-varying scales of variance. We show analytically that these two different variance predictions for the target are complementary and their combination leads to consistent and significant improvements in the sample efficiency and overall performance of the learning process.In summary, our contribution is threefold: | SAMPLE EFFICIENT DEEP REINFORCEMENT LEARN- ING VIA UNCERTAINTY ESTIMATION |
d211532692 | Imitation Learning (IL) is a popular paradigm for training agents to achieve complicated goals by leveraging expert behavior, rather than dealing with the hardships of designing a correct reward function. With the environment modeled as a Markov Decision Process (MDP), most of the existing IL algorithms are contingent on the availability of expert demonstrations in the same MDP as the one in which a new imitator policy is to be learned. This is uncharacteristic of many real-life scenarios where discrepancies between the expert and the imitator MDPs are common, especially in the transition dynamics function. Furthermore, obtaining expert actions may be costly or infeasible, making the recent trend towards state-only IL (where expert demonstrations constitute only states or observations) ever so promising. Building on recent adversarial imitation approaches that are motivated by the idea of divergence minimization, we present a new state-only IL algorithm in this paper. It divides the overall optimization objective into two subproblems by introducing an indirection step and solves the subproblems iteratively. We show that our algorithm is particularly effective when there is a transition dynamics mismatch between the expert and imitator MDPs, while the baseline IL methods suffer from performance degradation. To analyze this, we construct several interesting MDPs by modifying the configuration parameters for the MuJoCo locomotion tasks from OpenAI Gym 1 . | STATE-ONLY IMITATION WITH TRANSITION DYNAM- ICS MISMATCH |
d231847376 | We build on the recently proposed EigenGame that views eigendecomposition as a competitive game. EigenGame's updates are biased if computed using minibatches of data, which hinders convergence and more sophisticated parallelism in the stochastic setting. In this work, we propose an unbiased stochastic update that is asymptotically equivalent to EigenGame, enjoys greater parallelism allowing computation on datasets of larger sample sizes, and outperforms EigenGame in experiments. We present applications to finding the principal components of massive datasets and performing spectral clustering of graphs. We analyze and discuss our proposed update in the context of EigenGame and the shift in perspective from optimization to games. * denotes equal contribution. 1 µ signifies unbiased or unloaded and α denotes original. | EIGENGAME UNLOADED WHEN PLAYING GAMES IS BETTER THAN OPTIMIZING |
d252683779 | Conformal prediction is a distribution-free technique for establishing valid prediction intervals. Although conventionally people conduct conformal prediction in the output space, this is not the only possibility. In this paper, we propose feature conformal prediction, which extends the scope of conformal prediction to semantic feature spaces by leveraging the inductive bias of deep representation learning. From a theoretical perspective, we demonstrate that feature conformal prediction provably outperforms regular conformal prediction under mild assumptions. Our approach could be combined with not only vanilla conformal prediction, but also other adaptive conformal prediction methods. Apart from experiments on existing predictive inference benchmarks, we also demonstrate the state-of-the-art performance of the proposed methods on large-scale tasks such as ImageNet classification and Cityscapes image segmentation. The code is available at https://github.com/AlvinWen428/FeatureCP. * Equal Contribution. † Correspond to | Published as a conference paper at ICLR 2023 PREDICTIVE INFERENCE WITH FEATURE CONFORMAL PREDICTION |
d251649129 | For supervised learning with tabular data, decision tree ensembles produced via boosting techniques generally dominate real-world applications involving iid training/test sets. However for graph data where the iid assumption is violated due to structured relations between samples, it remains unclear how to best incorporate this structure within existing boosting pipelines. To this end, we propose a generalized framework for iterating boosting with graph propagation steps that share node/sample information across edges connecting related samples. Unlike previous efforts to integrate graph-based models with boosting, our approach is anchored in a principled meta loss function such that provable convergence can be guaranteed under relatively mild assumptions. Across a variety of non-iid graph datasets with tabular node features, our method achieves comparable or superior performance than both tabular and graph neural network models, as well as existing hybrid strategies that combine the two. Beyond producing better predictive performance than recently proposed graph models, our proposed techniques are easy to implement, computationally more efficient, and enjoy stronger theoretical guarantees (which make our results more reproducible). The source code is available at https://github.com. Attention-based graph neural network for semi-supervised learning. arXiv preprint arXiv:1803.03735, 2018. of tricks for node classification with graph neural networks. arXiv preprint arXiv:2103.13355, 2021. | Published as a conference paper at ICLR 2022 DOES YOUR GRAPH NEED A CONFIDENCE BOOST? CONVERGENT BOOSTED SMOOTHING ON GRAPHS WITH TABULAR NODE FEATURES |
d165163910 | Deep neural networks (DNNs) have great expressive power, which can even memorize samples with wrong labels. It is vitally important to reiterate robustness and generalization in DNNs against label corruption. To this end, this paper studies the 0-1 loss, which has a monotonic relationship with empirical adversary (reweighted) risk(Hu et al., 2018). Although the 0-1 loss has some robust properties, it is difficult to optimize. To efficiently optimize the 0-1 loss while keeping its robust properties, we propose a very simple and efficient loss, i.e. curriculum loss (CL). Our CL is a tighter upper bound of the 0-1 loss compared with conventional summation based surrogate losses. Moreover, CL can adaptively select samples for model training. As a result, our loss can be deemed as a novel perspective of curriculum sample selection strategy, which bridges a connection between curriculum learning and robust learning. Experimental results on benchmark datasets validate the robustness of the proposed loss. | Published as a conference paper at ICLR 2020 CURRICULUM LOSS: ROBUST LEARNING AND GEN- ERALIZATION AGAINST LABEL CORRUPTION |
d251979354 | Deep reinforcement learning agents are notoriously sample inefficient, which considerably limits their application to real-world problems. Recently, many model-based methods have been designed to address this issue, with learning in the imagination of a world model being one of the most prominent approaches. However, while virtually unlimited interaction with a simulated environment sounds appealing, the world model has to be accurate over extended periods of time. Motivated by the success of Transformers in sequence modeling tasks, we introduce IRIS, a data-efficient agent that learns in a world model composed of a discrete autoencoder and an autoregressive Transformer. With the equivalent of only two hours of gameplay in the Atari 100k benchmark, IRIS achieves a mean human normalized score of 1.046, and outperforms humans on 10 out of 26 games, setting a new state of the art for methods without lookahead search. To foster future research on Transformers and world models for sample-efficient reinforcement learning, we release our code and models at https://github.com/eloialonso/iris. , et al. Human-level control through deep reinforcement learning. Nature, 518(7540):529-533, 2015. Silver. Mastering atari, go, chess and shogi by planning with a learned model. Nature, 588(7839): 604-609, 2020. | Published as a conference paper at ICLR 2023 TRANSFORMERS ARE SAMPLE-EFFICIENT WORLD MODELS |
d246706127 | Physics-inspired neural networks (NNs), such as Hamiltonian or Lagrangian NNs, dramatically outperform other learned dynamics models by leveraging strong inductive biases. These models, however, are challenging to apply to many real world systems, such as those that don't conserve energy or contain contacts, a common setting for robotics and reinforcement learning. In this paper, we examine the inductive biases that make physics-inspired models successful in practice. We show that, contrary to conventional wisdom, the improved generalization of HNNs is the result of modeling acceleration directly and avoiding artificial complexity from the coordinate system, rather than symplectic structure or energy conservation. We show that by relaxing the inductive biases of these models, we can match or exceed performance on energy-conserving systems while dramatically improving performance on practical, non-conservative systems. We extend this approach to constructing transition models for common Mujoco environments, showing that our model can appropriately balance inductive biases with the flexibility required for model-based control. 25% 50% 75% 100% Performance + ODE Bias + Second-Order Bias + Symplectic Bias Hamiltonian NN Expectation 25% 50% 75% 100% Performance + ODE Bias + Second-Order Bias + Symplectic Bias Hamiltonian NN RealityFigure 1: The common perception in physics-informed machine learning is that increased performance is the result of complex biases. We find, however, that simpler implicit biases (such as second-order structure) often account for almost all of the improvement over baselines.Published as a conference paper at ICLR 2022Cranmer et al., 2020) share the same structure and inductive biases as HNNs, we focus on HNNs where energy conservation and symplecticity are more explicit.HNNs encode a number of inductive biases that help model physical systems: | Published as a conference paper at ICLR 2022 DECONSTRUCTING THE INDUCTIVE BIASES OF HAMILTONIAN NEURAL NETWORKS |
d246016131 | Class-conditioning offers a direct means to control a Generative Adversarial Network (GAN) based on a discrete input variable. While necessary in many applications, the additional information provided by the class labels could even be expected to benefit the training of the GAN itself. On the contrary, we observe that class-conditioning causes mode collapse in limited data settings, where unconditional learning leads to satisfactory generative ability. Motivated by this observation, we propose a training strategy for class-conditional GANs (cGANs) that effectively prevents the observed mode-collapse by leveraging unconditional learning. Our training strategy starts with an unconditional GAN and gradually injects the class conditioning into the generator and the objective function. The proposed method for training cGANs with limited data results not only in stable training but also in generating high-quality images, thanks to the early-stage exploitation of the shared information across classes. We analyze the observed mode collapse problem in comprehensive experiments on four datasets. Our approach demonstrates outstanding results compared with stateof-the-art methods and established baselines. The code is available at https: / | Published as a conference paper at ICLR 2022 COLLAPSE BY CONDITIONING: TRAINING CLASS- CONDITIONAL GANS WITH LIMITED DATA |
d250526398 | Hypergraph neural networks (HNNs) using neural networks to encode hypergraphs provide a promising way to model higher-order relations in data and further solve relevant prediction tasks built upon such higher-order relations. However, higher-order relations in practice contain complex patterns and are often highly irregular. So, it is often challenging to design an HNN that suffices to express those relations while keeping computational efficiency. Inspired by hypergraph diffusion algorithms, this work proposes a new HNN architecture named ED-HNN, which provably approximates any continuous equivariant hypergraph diffusion operators that can model a wide range of higher-order relations. ED-HNN can be implemented efficiently by combining star expansions of hypergraphs with standard message passing neural networks. ED-HNN further shows great superiority in processing heterophilic hypergraphs and constructing deep models. We evaluate ED-HNN for node classification on nine real-world hypergraph datasets. ED-HNN uniformly outperforms the best baselines over these nine datasets and achieves more than 2%↑ in prediction accuracy over four datasets therein. Our | Published as a conference paper at ICLR 2023 EQUIVARIANT HYPERGRAPH DIFFUSION NEURAL OP- ERATORS |
d211133302 | The information bottleneck principle provides an information-theoretic method for representation learning, by training an encoder to retain all information which is relevant for predicting the label while minimizing the amount of other, excess information in the representation. The original formulation, however, requires labeled data to identify the superfluous information. In this work, we extend this ability to the multi-view unsupervised setting, where two views of the same underlying entity are provided but the label is unknown. This enables us to identify superfluous information as that not shared by both views. A theoretical analysis leads to the definition of a new multi-view model that produces state-of-the-art results on the Sketchy dataset and label-limited versions of the MIR-Flickr dataset. We also extend our theory to the single-view setting by taking advantage of standard data augmentation techniques, empirically showing better generalization capabilities when compared to common unsupervised approaches for representation learning. | Published as a conference paper at ICLR 2020 LEARNING ROBUST REPRESENTATIONS VIA MULTI-VIEW INFORMATION BOTTLENECK |
d216077591 | Differential equations are frequently used in engineering domains, such as modeling and control of industrial systems, where safety and performance guarantees are of paramount importance. Traditional physics-based modeling approaches require domain expertise and are often difficult to tune or adapt to new systems. In this paper, we show how to model discrete ordinary differential equations (ODE) with algebraic nonlinearities as deep neural networks with varying degrees of prior knowledge. We derive the stability guarantees of the network layers based on the implicit constraints imposed on the weight's eigenvalues. Moreover, we show how to use barrier methods to generically handle additional inequality constraints. We demonstrate the prediction accuracy of learned neural ODEs evaluated on open-loop simulations compared to ground truth dynamics with bi-linear terms. | Constrained Neural Ordinary Differential Equations with Stability Guarantees |
d247475806 | As reinforcement learning (RL) has achieved near human-level performance in a variety of tasks, its robustness has raised great attention. While a vast body of research has explored test-time (evasion) attacks in RL and corresponding defenses, its robustness against training-time (poisoning) attacks remains largely unanswered. In this work, we focus on certifying the robustness of offline RL in the presence of poisoning attacks, where a subset of training trajectories could be arbitrarily manipulated. We propose the first certification framework, COPA, to certify the number of poisoning trajectories that can be tolerated regarding different certification criteria. Given the complex structure of RL, we propose two certification criteria: per-state action stability and cumulative reward bound. To further improve the certification, we propose new partition and aggregation protocols to train robust policies. We further prove that some of the proposed certification methods are theoretically tight and some are NP-Complete problems. We leverage COPA to certify three RL environments trained with different algorithms and conclude: (1) The proposed robust aggregation protocols such as temporal aggregation can significantly improve the certifications; (2) Our certifications for both per-state action stability and cumulative reward bound are efficient and tight;(3) The certification for different training algorithms and environments are different, implying their intrinsic robustness properties. All experimental results are available at https://copa-leaderboard.github.io.Published as a conference paper at ICLR 2022 method that is able to provide practically computable certified robustness against poisoning attacks. In this paper, we tackle this problem by proposing the first framework of Certifying robust policies for general offline RL against poisoning attacks (COPA).Certification Criteria. One critical challenge in certifying robustness for offline RL is the certification criteria, since the prediction consistency is no longer the only goal as in classification. We propose two criteria based on the properties of RL: per-state action stability and cumulative reward bound. The former guarantees that at a specific time, the policy learned with COPA will predict the same action before and after attacks under certain conditions. This is important for guaranteeing the safety of the policy at critical states, e.g., braking when seeing pedestrians. For cumulative reward bound, a lower bound of the cumulative reward for the policy learned with COPA is guaranteed under certain poisoning conditions. This directly guarantees the worst-case overall performance.COPA Framework. COPA is composed of two components: policy partition and aggregation protocol and robustness certification method. We propose three policy partition aggregation protocols: PARL (Per-State Partition Aggregation), TPARL (Temporal Partition Aggregation), and DPARL (Dynamic Temporal Partition Aggregation), and propose certification methods for each of them corresponding to both proposed certification criteria. In addition, for per-state action stability, we prove that our certifications for PARL and TPARL are theoretically tight. For cumulative reward bound, we propose an adaptive search algorithm, where we compute the possible action set for each state under certain poisoning conditions. Concretely, we propose a novel method to compute the precise action set for PARL and efficient algorithms to compute a superset of the possible action set which leads to sound certification for TPARL and DPARL. We further prove that for PARL our certification is theoretically tight, for TPARL the theoretically tight certification is NP-complete, and for DPARL it is open whether theoretically tight certification exists.Technical Contributions. We take the first step towards certifying the robustness of offline RL against poisoning attacks, and we make contributions on both theoretical and practical fronts.• We abstract and formulate the robustness certification for offline RL against poisoning attacks, and we propose two certification criteria: per-state action stability and cumulative reward bound. • We propose the first framework COPA for certifying robustness of offline RL against poisoning attacks. COPA includes novel policy aggregation protocols and certification methods. • We prove the tightness of the proposed certification methods for the aggregation protocol PARL.We also prove the computational hardness of the certification for TPARL. • We conduct thorough experimental evaluation for COPA on different RL environments with three offline RL algorithms, demonstrating the effectiveness of COPA, together with several interesting findings.RELATED WORKPoisoning attacks (Nelson et al., 2008; Diakonikolas et al., 2016) are critical threats in machine learning, which are claimed to be more concerning than other threats (Kumar et al., 2020). Poisoning attacks widely exist in classification (Schwarzschild et al., 2021), and both empirical defenses (Liu et al., 2018; Chacon et al., 2019; Peri et al., 2020; Steinhardt et al., 2017) and certified defenses (Weber et al., 2020; Jia et al., 2020; Levine & Feizi, 2021) have been proposed. After Kiourti et al. (2020) show the existence of effective backdoor poisoning attacks in RL, a recent work theoretically and empirically validates the existence of reward poisoning in online RL (Zhang et al., 2020b). Furthermore, Zhang et al. (2021) theoretically prove that the offline RL is more difficult to be robustified against poisoning than online RL considering linear MDP. From the defense side, Zhang et al. (2021) propose robust variants of the Least-Square Value Iteration algorithm that provides probabilistic robustness guarantees under linear MDP assumption. In addition, Robust RL against reward poisoning is studied in Banihashem et al. (2021), but robust RL against general poisoning is less explored. In this background, we aim to provide the certified robustness for general offline RL algorithms against poisoning attacks, which is the first work that achieves the goal. We discuss broader related work in Appendix I.CERTIFICATION CRITERIA OF COPAIn this section, we propose two robustness certification criteria for offline RL against general poisoning attacks: per-state action stability and cumulative reward bound.2 Published as a conference paper at ICLR 2022Offline RL. We model the RL environment by an episodic finite-horizon Markov decision process (MDP) E = (S, A, R, P, H, d 0 ), where S is the set of states, A is the set of discrete actions, R : S × A → R is the reward function, P : S × A → P(S) is the stochastic transition function with P(·) defining the set of probability measures, H is the time horizon, and d 0 ∈ P(S) is the distribution of the initial state. At time step t, the RL agent is at state s t ∈ S. After choosing action a t ∈ A, the agent transitions to the next state s t+1 ∼ P (s t , a t ) and receives reward r t = R(s t , a t ).After H time steps, the cumulative reward J = H−1 t=0 r t . We denote a consecutive sequence of all states between time step l and r as s l:r := [s l , s l+1 , . . . , s r ].Here we focus on offline RL, for which the threat of poisoning attacks is practical and more challenging to deal with (Zhang et al., 2021). Concretely, in offline RL, a training datasetconsists of logged trajectories, where each trajectory τ = {(s j , r j , a j , s j )} l j=1 ∈ (S × A × R × S) l consists of multiple tuples denoting the transitions (i.e., starting from state s j , taking the action a j , receiving reward r j , and transitioning to the next state s j ).Poisoning Attacks. Training dataset D can be poisoned in the following manner. For each trajectory τ ∈ D, the adversary is allowed to replace it with an arbitrary trajectory τ , generating a manipulated dataset D. We denote D D = (D\ D) ( D\D) as the symmetric difference between two datasets D and D. For instance, adding or removing one trajectory causes a symmetric difference of magnitude 1, while replacing one trajectory with a new one leads to a symmetric difference of magnitude 2. We refer to the size of the symmetric difference as the poisoning size.Certification Goal. To provide the robustness certification against poisoning attacks introduced above, we aim to certify the test-time performance of the trained policy in a clean environment. Specifically, in the training phase, the RL training algorithm and our aggregation protocol can be jointly modeled by M : D → (S → A) which provides an aggregated policy, where S denotes the set of all consecutive state sequences. Our goal is to provide robustness certification for the poisoned aggregated policyπ = M( D), given bounded poisoning size (i.e., |D D| ≤ K).Robustness Certification Criteria: Per-State Action Stability. We first aim to certify the robustness of the poisoned policy in terms of the stability of per-state action during test time.Definition 1 (Robustness Certification for Per-State Action Stability). Given a clean dataset D, we define the robustness certification for per-state action stability as that for any D satisfying |D D| ≤ K, the action predictions of the poisoned and clean policies for the state (or state sequence) s are the same, i.e.,π = M( D), π = M(D),π(s) = π(s), under the the tolerable poisoning threshold K.In an episode, we denote the tolerable poisoning threshold for the state at step t by K t . | COPA: CERTIFYING ROBUST POLICIES FOR OF- FLINE REINFORCEMENT LEARNING AGAINST POISON- ING ATTACKS |
d246652076 | Though remarkable progress has been achieved in various vision tasks, deep neural networks still suffer obvious performance degradation when tested in out-ofdistribution scenarios. We argue that the feature statistics (mean and standard deviation), which carry the domain characteristics of the training data, can be properly manipulated to improve the generalization ability of deep learning models. Common methods often consider the feature statistics as deterministic values measured from the learned features and do not explicitly consider the uncertain statistics discrepancy caused by potential domain shifts during testing. In this paper, we improve the network generalization ability by modeling the uncertainty of domain shifts with synthesized feature statistics during training. Specifically, we hypothesize that the feature statistic, after considering the potential uncertainties, follows a multivariate Gaussian distribution. Hence, each feature statistic is no longer a deterministic value, but a probabilistic point with diverse distribution possibilities. With the uncertain feature statistics, the models can be trained to alleviate the domain perturbations and achieve better robustness against potential domain shifts. Our method can be readily integrated into networks without additional parameters. Extensive experiments demonstrate that our proposed method consistently improves the network generalization ability on multiple vision tasks, including image classification, semantic segmentation, and instance retrieval. The code can be available at https://github.com/lixiaotong97/DSU. | Published as a conference paper at ICLR 2022 UNCERTAINTY MODELING FOR OUT-OF-DISTRIBUTION GENERALIZATION |
d247362438 | Wide neural networks with linear output layer have been shown to be near-linear, and to have near-constant neural tangent kernel (NTK), in a region containing the optimization path of gradient descent. These findings seem counter-intuitive since in general neural networks are highly complex models. Why does a linear structure emerge when the networks become wide? In this work, we provide a new perspective on this "transition to linearity" by considering a neural network as an assembly model recursively built from a set of sub-models corresponding to individual neurons. In this view, we show that the linearity of wide neural networks is, in fact, an emerging property of assembling a large number of diverse "weak" sub-models, none of which dominate the assembly. | TRANSITION TO LINEARITY OF WIDE NEURAL NET- WORKS IS AN EMERGING PROPERTY OF ASSEMBLING WEAK MODELS |
d211044132 | We propose a novel node embedding of directed graphs to statistical manifolds, which is based on a global minimization of pairwise relative entropy and graph geodesics in a non-linear way. Each node is encoded with a probability density function over a measurable space. Furthermore, we analyze the connection between the geometrical properties of such embedding and their efficient learning procedure. Extensive experiments show that our proposed embedding is better in preserving the global geodesic information of graphs, as well as outperforming existing embedding models on directed graphs in a variety of evaluation metrics, in an unsupervised setting. | LOW-DIMENSIONAL STATISTICAL MANIFOLD EMBED- DING OF DIRECTED GRAPHS |
d251252882 | Recent large-scale text-driven synthesis models have attracted much attention thanks to their remarkable capabilities of generating highly diverse images that follow given text prompts. Such text-based synthesis methods are particularly appealing to humans who are used to verbally describe their intent. Therefore, it is only natural to extend the text-driven image synthesis to text-driven image editing. Editing is challenging for these generative models, since an innate property of an editing technique is to preserve most of the original image, while in the text-based models, even a small modification of the text prompt often leads to a completely different outcome. State-of-the-art methods mitigate this by requiring the users to provide a spatial mask to localize the edit, hence, ignoring the original structure and content within the masked region. In this paper, we pursue an intuitive prompt-toprompt editing framework, where the edits are controlled by text only. To this end, we analyze a text-conditioned model in depth and observe that the cross-attention layers are the key to controlling the relation between the spatial layout of the image to each word in the prompt. With this observation, we present several applications which monitor the image synthesis by editing the textual prompt only. This includes localized editing by replacing a word, global editing by adding a specification, and even delicately controlling the extent to which a word is reflected in the image. We present our results over diverse images and prompts, demonstrating high-quality synthesis and fidelity to the edited prompts. | Prompt-to-Prompt Image Editing with Cross Attention Control |
d786898 | Recently, image representation built upon Convolutional Neural Network (CNN) has been shown to provide effective descriptors for image search, outperforming pre-CNN features as short-vector representations. Yet such models are not compatible with geometry-aware re-ranking methods and still outperformed, on some particular object retrieval benchmarks, by traditional image search systems relying on precise descriptor matching, geometric re-ranking, or query expansion. This work revisits both retrieval stages, namely initial search and re-ranking, by employing the same primitive information derived from the CNN. We build compact feature vectors that encode several image regions without the need to feed multiple inputs to the network. Furthermore, we extend integral images to handle max-pooling on convolutional layer activations, allowing us to efficiently localize matching objects. The resulting bounding box is finally used for image reranking. As a result, this paper significantly improves existing CNN-based recognition pipeline: We report for the first time results competing with traditional methods on the challenging Oxford5k and Paris6k datasets. * Research partially conducted while G. Tolias and H. Jégou were at Inria. We would like to thank Florent Perronnin for his valuable feedback. | PARTICULAR OBJECT RETRIEVAL WITH INTEGRAL MAX-POOLING OF CNN ACTIVATIONS |
d231985703 | Mixed-precision quantization can potentially achieve the optimal tradeoff between performance and compression rate of deep neural networks, and thus, have been widely investigated. However, it lacks a systematic method to determine the exact quantization scheme. Previous methods either examine only a small manuallydesigned search space or utilize a cumbersome neural architecture search to explore the vast search space. These approaches cannot lead to an optimal quantization scheme efficiently. This work proposes bit-level sparsity quantization (BSQ) to tackle the mixed-precision quantization from a new angle of inducing bit-level sparsity. We consider each bit of quantized weights as an independent trainable variable and introduce a differentiable bit-sparsity regularizer. BSQ can induce all-zero bits across a group of weight elements and realize the dynamic precision reduction, leading to a mixed-precision quantization scheme of the original model. Our method enables the exploration of the full mixed-precision space with a single gradient-based optimization process, with only one hyperparameter to tradeoff the performance and compression. BSQ achieves both higher accuracy and higher bit reduction on various model architectures on the CIFAR-10 and ImageNet datasets comparing to previous methods. | Published as a conference paper at ICLR 2021 BSQ: EXPLORING BIT-LEVEL SPARSITY FOR MIXED- PRECISION NEURAL NETWORK QUANTIZATION |
d210942708 | Recent research developing neural network architectures with external memory have often used the benchmark bAbI question and answering dataset which provides a challenging number of tasks requiring reasoning. Here we employed a classic associative inference task from the memory-based reasoning neuroscience literature in order to more carefully probe the reasoning capacity of existing memoryaugmented architectures. This task is thought to capture the essence of reasoningthe appreciation of distant relationships among elements distributed across multiple facts or memories. Surprisingly, we found that current architectures struggle to reason over long distance associations. Similar results were obtained on a more complex task involving finding the shortest path between nodes in a path. We therefore developed MEMO, an architecture endowed with the capacity to reason over longer distances. This was accomplished with the addition of two novel components. First, it introduces a separation between memories/facts stored in external memory and the items that comprise these facts in external memory. Second, it makes use of an adaptive retrieval mechanism, allowing a variable number of 'memory hops' before the answer is produced. MEMO is capable of solving our novel reasoning tasks, as well as match state of the art results in bAbI. * . Big-loop recurrence within the hippocampal system supports integration of information across episodes. Neuron, 99(6):1342-1354, 2018. . Learning deep generative models of graphs. arXiv preprint arXiv:1803.03324, 2018. Christos Louizos, Max Welling, and Diederik P Kingma. Learning sparse neural networks through l_0 regularization. arXiv preprint arXiv:1712.01312, 2017. . A simple neural network module for relational reasoning. In Advances in neural information processing systems, pp. 4967-4976, 2017. graph neural network model. Norman. Complementary learning systems within the hippocampus: a neural network modelling approach to reconciling episodic memory with statistical learning. | MEMO: A DEEP NETWORK FOR FLEXIBLE COMBINA- TION OF EPISODIC MEMORIES |
d211076166 | Identifying when to give treatments to patients and how to select among multiple treatments over time are important medical problems with a few existing solutions.In this paper, we introduce the Counterfactual Recurrent Network (CRN), a novel sequence-to-sequence model that leverages the increasingly available patient observational data to estimate treatment effects over time and answer such medical questions.To handle the bias from time-varying confounders, covariates affecting the treatment assignment policy in the observational data, CRN uses domain adversarial training to build balancing representations of the patient history.At each timestep, CRN constructs a treatment invariant representation which removes the association between patient history and treatment assignments and thus can be reliably used for making counterfactual predictions.On a simulated model of tumour growth, with varying degree of time-dependent confounding, we show how our model achieves lower error in estimating counterfactuals and in choosing the correct treatment and timing of treatment than current state-of-the-art methods. | |
d252683961 | Lighter and faster image restoration (IR) models are crucial for the deployment on resource-limited devices. Binary neural network (BNN), one of the most promising model compression methods, can dramatically reduce the computations and parameters of full-precision convolutional neural networks (CNN). However, there are different properties between BNN and full-precision CNN, and we can hardly use the experience of designing CNN to develop BNN. In this study, we reconsider components in binary convolution, such as residual connection, Batch-Norm, activation function, and structure, for IR tasks. We conduct systematic analyses to explain each component's role in binary convolution and discuss the pitfalls. Specifically, we find that residual connection can reduce the information loss caused by binarization; BatchNorm can solve the value range gap between residual connection and binary convolution; The position of the activation function dramatically affects the performance of BNN. Based on our findings and analyses, we design a simple yet efficient basic binary convolution unit (BBCU). Furthermore, we divide IR networks into four parts and specially design variants of BBCU for each part to explore the benefit of binarizing these parts. We conduct experiments on different IR tasks, and our BBCU significantly outperforms other BNNs and lightweight models, which shows that BBCU can serve as a basic unit for binarized IR networks. The code is available at https://github.com/ | Published as a conference paper at ICLR 2023 BASIC BINARY CONVOLUTION UNIT FOR BINARIZED IMAGE RESTORATION NETWORK |
d3284333 | Deep learning software demands reliability and performance. However, many of the existing deep learning frameworks are software libraries that act as an unsafe DSL in Python and a computation graph interpreter. We present DLVM, a design and implementation of a compiler infrastructure with a linear algebra intermediate representation, algorithmic differentiation by adjoint code generation, domainspecific optimizations and a code generator targeting GPU via LLVM. Designed as a modern compiler infrastructure inspired by LLVM, DLVM is more modular and more generic than existing deep learning compiler frameworks, and supports tensor DSLs with high expressivity. With our prototypical staged DSL embedded in Swift, we argue that the DLVM system enables a form of modular, safe and performant frameworks for deep learning. | Workshop track -ICLR 2018 DLVM: A MODERN COMPILER INFRASTRUCTURE FOR DEEP LEARNING SYSTEMS |
d54447167 | We study adversarial robustness of neural networks from a margin maximization perspective, where margins are defined as the distances from inputs to a classifier's decision boundary. Our study shows that maximizing margins can be achieved by minimizing the adversarial loss on the decision boundary at the "shortest successful perturbation", demonstrating a close connection between adversarial losses and the margins. We propose Max-Margin Adversarial (MMA) training to directly maximize the margins to achieve adversarial robustness. Instead of adversarial training with a fixed , MMA offers an improvement by enabling adaptive selection of the "correct" as the margin individually for each data point. In addition, we rigorously analyze adversarial training with the perspective of margin maximization, and provide an alternative interpretation for adversarial training, maximizing either a lower or an upper bound of the margins. Our experiments empirically confirm our theory and demonstrate MMA training's efficacy on the MNIST and CIFAR10 datasets w.r.t. ∞ and 2 robustness. Code and models are available at https://github.com/BorealisAI/mma_training. | Published as a conference paper at ICLR 2020 MMA TRAINING: DIRECT INPUT SPACE MARGIN MAXIMIZATION THROUGH ADVERSARIAL TRAINING |
d231718715 | Federated learning (FL) is a distributed machine learning architecture that leverages a large number of workers to jointly learn a model with decentralized data. FL has received increasing attention in recent years thanks to its data privacy protection, communication efficiency and a linear speedup for convergence in training (i.e., convergence performance increases linearly with respect to the number of workers). However, existing studies on linear speedup for convergence are only limited to the assumptions of i.i.d. datasets across workers and/or full worker participation, both of which rarely hold in practice. So far, it remains an open question whether or not the linear speedup for convergence is achievable under non-i.i.d. datasets with partial worker participation in FL. In this paper, we show that the answer is affirmative. Specifically, we show that the federated averaging (FedAvg) algorithm (with two-sided learning rates) on non-i.i.d. datasets in non-convex settings achieves a convergence rate O( 1 √ mKT + 1 T ) for full worker participation and a convergence rate O( 1 √ nKT + 1 T ) for partial worker participation, where K is the number of local steps, T is the number of total communication rounds, m is the total worker number and n is the worker number in one communication round if for partial worker participation. Our results also reveal that the local steps in FL could help the convergence and show that the maximum number of local steps can be improved to T /m. We conduct extensive experiments on MNIST and CIFAR-10 to verify our theoretical results.Published as a conference paper at ICLR 2021• Time-varying partial worker participation (systems non-stationarity): With the flexibility for workers' participation in many scenarios (particularly in mobile edge computing), workers may randomly join or leave the FL system at will, thus rendering the active worker set stochastic and time-varying across communication rounds. Hence, it is often infeasible to wait for all workers' responses as in traditional distributed learning, since inactive workers or stragglers will significantly slow down the whole training process. As a result, only a subset of the workers may be chosen by the server in each communication round, i.e., partial worker participation.In recent years, the Federated Averaging method (FedAvg) and its variants(McMahan et al., 2016; Li et al., 2018; Hsu et al., 2019; Karimireddy et al., 2019; Wang et al., 2019a) have emerged as a prevailing approach for FL. Similar to the traditional distributed learning, FedAvg leverages local computation at each worker and employs a centralized parameter server to aggregate and update the model parameters. The unique feature of FedAvg is that each worker runs multiple local stochastic gradient descent (SGD) steps rather than just one step as in traditional distributed learning between two consecutive communication rounds. For i.i.d. datasets and the full worker participation setting, Stich (2018) and Yu et al. (2019b) proposed two variants of FedAvg that achieve a convergence rate of O( mK T + 1 √ mKT ) with a bounded gradient assumption for both strongly convex and nonconvex problems, where m is the number of workers, K is the local update steps and T is the total communication rounds. Wang & Joshi (2018) and Stich & Karimireddy (2019) further proposed improved FedAvg algorithms to achieve a O( m T + 1 √ mKT ) rate without bounded gradient assumption. Notably, for a sufficiently large T , the above rates become O( 1 √ mKT ) 1 , which implies a linear speedup with respect to the number of workers. 2 This linear speedup is highly desirable for an FL algorithm because the algorithm is able to effectively leverage the massive parallelism in a large FL system. However, with non-i.i.d. datasets and partial worker participation in FL, a fundamental open question arises: Can we still achieve the same linear speedup for convergence, i.e., O( 1 √ mKT ), with non-i.i.d. datasets and under either full or partial worker participation? • In order to achieve a linear speedup, i.e., a convergence rate O( 1 √ mKT ), we show that the number of local updates K can be as large as T /m, which improves the T 1/3 /m result previously shown in Yu et al. (2019a) and Karimireddy et al. (2019). As shown later in the communication complexity comparison inTable 1, a larger number of local steps implies relatively fewer communication rounds, thus less communication overhead. Interestingly, our results also indicate that the number of local updates K does not hurt but rather help the convergence with a proper learning rates choice. This overcomes the limitation as suggested in Li et al. (2019b) that local SGD steps might slow down the convergence (O( K T ) for strongly convex case). This result also reveals new insights on the relationship between the number of local steps and learning rate. 1 This rate also matches the convergence rate order of parallel SGD in conventional distributed learning. 2 To attain accuracy for an algorithm, it needs to take O( 1 2 ) steps with a convergence rate O( 1 √ T ), while needing O( 1 m 2 ) steps if the convergence rate is O( 1 √ mT ) (the hidden constant in Big-O is the same). In this sense, one achieves a linear speedup with respect to the number of workers. | Published as a conference paper at ICLR 2021 ACHIEVING LINEAR SPEEDUP WITH PARTIAL WORKER PARTICIPATION IN NON-IID FEDERATED LEARNING |
d248006105 | Supervised federated learning (FL) enables multiple clients to share the trained model without sharing their labeled data. However, potential clients might even be reluctant to label their own data, which could limit the applicability of FL in practice. In this paper, we show the possibility of unsupervised FL whose model is still a classifier for predicting class labels, if the class-prior probabilities are shifted while the class-conditional distributions are shared among the unlabeled data owned by the clients. We propose federation of unsupervised learning (FedUL), where the unlabeled data are transformed into surrogate labeled data for each of the clients, a modified model is trained by supervised FL, and the wanted model is recovered from the modified model. FedUL is a very general solution to unsupervised FL: it is compatible with many supervised FL methods, and the recovery of the wanted model can be theoretically guaranteed as if the data have been labeled. | Published as a conference paper at ICLR 2022 FEDERATED LEARNING FROM ONLY UNLABELED DATA WITH CLASS-CONDITIONAL-SHARING CLIENTS |
d257353246 | Many recent works have shown trainability plays a central role in neural network pruning -unattended broken trainability can lead to severe under-performance and unintentionally amplify the effect of retraining learning rate, resulting in biased (or even misinterpreted) benchmark results. This paper introduces trainability preserving pruning (TPP), a scalable method to preserve network trainability against pruning, aiming for improved pruning performance and being more robust to retraining hyper-parameters (e.g., learning rate). Specifically, we propose to penalize the gram matrix of convolutional filters to decorrelate the pruned filters from the retained filters. In addition to the convolutional layers, per the spirit of preserving the trainability of the whole network, we also propose to regularize the batch normalization parameters (scale and bias). Empirical studies on linear MLP networks show that TPP can perform on par with the oracle trainability recovery scheme. On nonlinear ConvNets (ResNet56/VGG19) on CIFAR10/100, TPP outperforms the other counterpart approaches by an obvious margin. Moreover, results on ImageNet-1K with ResNets suggest that TPP consistently performs more favorably against other top-performing structured pruning approaches. Code: https://github.com/MingSun-Tse/TPP. | Published as a conference paper at ICLR 2023 TRAINABILITY PRESERVING NEURAL PRUNING |
d4606753 | Synthesizing user-intended programs from a small number of input-output examples is a challenging problem with several important applications like spreadsheet manipulation, data wrangling and code refactoring. Existing synthesis systems either completely rely on deductive logic techniques that are extensively handengineered or on purely statistical models that need massive amounts of data, and in general fail to provide real-time synthesis on challenging benchmarks. In this work, we propose Neural Guided Deductive Search (NGDS), a hybrid synthesis technique that combines the best of both symbolic logic techniques and statistical models. Thus, it produces programs that satisfy the provided specifications by construction and generalize well on unseen examples, similar to data-driven systems. Our technique effectively utilizes the deductive search framework to reduce the learning problem of the neural component to a simple supervised learning setup. Further, this allows us to both train on sparingly available real-world data and still leverage powerful recurrent neural network encoders. We demonstrate the effectiveness of our method by evaluating on real-world customer scenarios by synthesizing accurate programs with up to 12× speed-up compared to state-ofthe-art systems. | NEURAL-GUIDED DEDUCTIVE SEARCH FOR REAL- TIME PROGRAM SYNTHESIS FROM EXAMPLES |
d221703133 | We study worst-case guarantees on the expected return of xed-dataset policy optimization algorithms. Our core contribution is a uni ed conceptual and mathematical framework for the study of algorithms in this regime.is analysis reveals that for naïve approaches, the possibility of erroneous value overestimation leads to a di cultto-satisfy requirement: in order to guarantee that we select a policy which is near-optimal, we may need the dataset to be informative of the value of every policy. To avoid this, algorithms can follow the pessimism principle, which states that we should choose the policy which acts optimally in the worst possible world. We show why pessimistic algorithms can achieve good performance even when the dataset is not informative of every policy, and derive families of algorithms which follow this principle. ese theoretical ndings are validated by experiments on a tabular gridworld, and deep learning experiments on four MinAtar environments. * Correspondence to: jacobbuckman@gmail.com 1 We use the term xed-dataset policy optimization to emphasize the computational procedure; this se ing has also been referred to as batch RL [5, 20] and more recently, o ine RL[23]. 2 e optimism principle states that we should select a policy with high best-case value. | e Importance of Pessimism in Fixed-Dataset Policy Optimization |
d6990380 | Figure 1: Our approach generates photorealistic output for various "incomplete" signals such as a low resolution image, a surface normal map, and edges/boundaries for human faces, cats, dogs, shoes, and handbags. Importantly, our approach can easily generate multiple outputs for a given input which was not possible in previous approaches[22]due to mode-collapse problem. Best viewed in electronic format.AbstractWe present a simple nearest-neighbor (NN) approach that synthesizes high-frequency photorealistic images from an "incomplete" signal such as a low-resolution image, a surface normal map, or edges. Current state-of-the-art deep generative models designed for such conditional image synthesis lack two important things: (1) they are unable to generate a large set of diverse outputs, due to the mode collapse problem. (2) they are not interpretable, making it difficult to control the synthesized output. We demonstrate that NN approaches potentially address such limitations, but suffer in accuracy on small datasets. We design a simple pipeline that combines the best of both worlds: the first stage uses a convolutional neural network (CNN) to maps the input to a (overly-smoothed) image, and the second stage uses a pixel-wise nearest neighbor method to map the smoothed output to multiple high-quality, highfrequency outputs in a controllable manner. We demonstrate our approach for various input modalities, and for various domains ranging from human faces to cats-and-dogs to shoes and handbags. | PixelNN: Example-based Image Synthesis 12x12 Input (x8) Our Approach (a) Low-Resolution to High-Resolution Surface Normal Map Our Approach (b) Normals-to-RGB Edges Our Approach (c) Edges-to-RGB (d) Edges-to-RGB (Multiple Outputs) (e) Normals-to-RGB (Multiple Outputs) (d) Edges-to-Shoes (Multiple Outputs) (e) Edges-to-Handbags (Multiple Outputs) |
d252383259 | The separation between training and deployment of machine learning models implies that not all scenarios encountered in deployment can be anticipated during training, and therefore relying solely on advancements in training has its limits. Out-of-distribution (OOD) detection is an important area that stress-tests a model's ability to handle unseen situations: Do models know when they don't know? Existing OOD detection methods either incur extra training steps, additional data or make nontrivial modifications to the trained network. In contrast, in this work, we propose an extremely simple, post-hoc, on-the-fly activation shaping method, ASH, where a large portion (e.g. 90%) of a sample's activation at a late layer is removed, and the rest (e.g. 10%) simplified or lightly adjusted. The shaping is applied at inference time, and does not require any statistics calculated from training data. Experiments show that such a simple treatment enhances in-distribution and out-ofdistribution distinction so as to allow state-of-the-art OOD detection on ImageNet, and does not noticeably deteriorate the in-distribution accuracy. Video, animation and code can be found at: https://andrijazz.github.io/ash. | Published as a conference paper at ICLR 2023 EXTREMELY SIMPLE ACTIVATION SHAPING FOR OUT- OF-DISTRIBUTION DETECTION |
d3433237 | It is widely believed that the success of deep convolutional networks is based on progressively discarding uninformative variability about the input with respect to the problem at hand. This is supported empirically by the difficulty of recovering images from their hidden representations, in most commonly used network architectures. In this paper we show via a one-to-one mapping that this loss of information is not a necessary condition to learn representations that generalize well on complicated problems, such as ImageNet. Via a cascade of homeomorphic layers, we build the i-RevNet, a network that can be fully inverted up to the final projection onto the classes, i.e. no information is discarded. Building an invertible architecture is difficult, for one, because the local inversion is ill-conditioned, we overcome this by providing an explicit inverse. An analysis of i-RevNets learned representations suggests an alternative explanation for the success of deep networks by a progressive contraction and linear separation with depth. To shed light on the nature of the model learned by the i-RevNet we reconstruct linear interpolations between natural image representations.A CNN may be very effective in classifying images of all sorts(He et al., 2016;Krizhevsky et al., 2012), but the cascade of linear and nonlinear operators reveals little about the contribution of the internal representation to the classification. The learning process is characterized by a steady reduction of large amounts of uninformative variability in the images while simultaneously revealing the essence of the visual class. It is widely believed that this process is based on progressively discarding uninformative variability about the input with respect to the problem at hand(Dosovitskiy & Brox, 2016;Mahendran & Vedaldi, 2016;Shwartz-Ziv & Tishby, 2017;Achille & Soatto, 2017). However, the extent to which information is discarded is lost somewhere in the intermediate nonlinear processing steps. In this paper, we aim to provide insight into the variability reduction process by proposing an invertible convolutional network, that does not discard any information about the input.The difficulty to recover images from their hidden representations is found in many commonly used network architectures(Dosovitskiy & Brox, 2016;Mahendran & Vedaldi, 2016). This poses the question if a substantial loss of information is necessary for successful classification. We show information does not have to be discarded. By using homeomorphic layers, the invariance can be built only at the very last layer via a projection.InShwartz-Ziv & Tishby (2017), minimal sufficient statistics are proposed as a candidate to explain the reduction of variability. Tishby & Zaslavsky (2015) introduces the information bottleneck principle which states that an optimal representation must reduce the mutual information between an input and its representation to reduce as much uninformative variability as possible. At the same time, the network should maximize the mutual information between the desired output and its representation to effectively preserve each class from collapsing onto other classes. | Published as a conference paper at ICLR 2018 i-REVNET: DEEP INVERTIBLE NETWORKS |
d258079386 | Modern image retrieval methods typically rely on fine-tuning pre-trained encoders to extract image-level descriptors. However, the most widely used models are pre-trained on ImageNet-1K with limited classes. The pre-trained feature representation is therefore not universal enough to generalize well to the diverse open-world classes. In this paper, we first cluster the large-scale LAION 400M dataset into one million pseudo classes based on the joint textual and visual features extracted by the CLIP model. Due to the confusion of label granularity, the automatically clustered dataset inevitably contains heavy inter-class conflict. To alleviate such conflict, we randomly select partial inter-class prototypes to construct the margin-based softmax loss. To further enhance the low-dimensional feature representation, we randomly select partial feature dimensions when calculating the similarities between embeddings and class-wise prototypes. The dual random partial selections are with respect to the class dimension and the feature dimension of the prototype matrix, making the classification conflict-robust and the feature embedding compact. Our method significantly outperforms state-of-the-art unsupervised and supervised image retrieval approaches on multiple benchmarks. The code and pre-trained models are released to facilitate future research https. Eurosat: A novel dataset and deep learning benchmark for land use and land cover classification. | Published as a conference paper at ICLR 2023 UNICOM: UNIVERSAL AND COMPACT REPRESENTATION LEARNING FOR IMAGE RETRIEVAL |
d6017868 | To address memory and computation resource limitations for hardware-oriented acceleration of deep convolutional neural networks (CNNs), we present a computation flow, stacked filters stationary flow (SFS), and a corresponding data encoding format, relative indexed compressed sparse filter format (CSF), to make the best of data sparsity, and simplify data handling at execution time. And we also propose a three dimensional Single Instruction Multiple Data (3D-SIMD) processor architecture which takes full advantage of these two features. Comparing with the state-of-the-art result (Han et al., 2016b), our method achieved 1.11× improvement in reducing the storage required by AlexNet, and 1.09× improvement in reducing the storage required by SqueezeNet, without loss of accuracy on the ImageNet dataset. Moreover, using this approach, chip area for logics handling irregular sparse data access can be saved. | STACKED FILTERS STATIONARY FLOW FOR HARDWARE-ORIENTED ACCELERATION OF DEEP CONVOLUTIONAL NEURAL NETWORKS |
d255393759 | Many recent works on understanding deep learning try to quantify how much individual data instances influence the optimization and generalization of a model. Such attempts reveal characteristics and importance of individual instances, which may provide useful information in diagnosing and improving deep learning. However, most of the existing works on data valuation require actual training of a model, which often demands high-computational cost. In this paper, we provide a training-free data valuation score, called complexity-gap score, which is a datacentric score to quantify the influence of individual instances in generalization of two-layer overparameterized neural networks. The proposed score can quantify irregularity of the instances and measure how much each data instance contributes in the total movement of the network parameters during training. We theoretically analyze and empirically demonstrate the effectiveness of the complexity-gap score in finding 'irregular or mislabeled' data instances, and also provide applications of the score in analyzing datasets and diagnosing training dynamics. Our code is publicly available at httpsPublished as a conference paper at ICLR 2023 two-layer neural networks trained by gradient descent. Different from that work, where the complexity of the full dataset was of main concern, our focus is on decomposing the effect of individual data instances in the training, and thus we newly introduce a complexity gap score (CG-score). We theoretically analyze and empirically demonstrate that the CG-score can quantify 'irregularity' of instances within each class, and thus can be used in identifying atypical examples, either due to the inherent irregularity of the instance or mislabeled classification. We also demonstrate that the proposed score has a close relation to 'learning difficulty' of the instances by analyzing the training dynamics of data instances. Our key contributions are as below:• Training-free data valuation: Different from previous methods for data valuation, most of which leverage the information from training itself, we provide a training-free data valuation score, CG-score, which is the data-centric score to quantify the effect of individual data instances in optimization and generalization of neural networks. • Geometric interpretation: We provide analysis that the CG-score can measure irregularity of each instance, i.e., it measures the average 'similarity' of an instance to the instances of the same class and the average 'dissimilarity' to the instances of different classes. • Effectiveness of the score: We empirically demonstrate the effectiveness of the CG-score in data valuation. We show that pruning data instances with small CG-score does not significantly degrade the generalization capability of a model, e.g., for CIFAR-10 we can prune 40% of the data with less than 1% of drop in test accuracy. Our scoring method is especially useful in data pruning, since different from other scores, which require the training with the full dataset, our method does not require any training of a model. • Application of the score: We provide potential applications of the CG-score in analyzing datasets and training dynamics. We analyze the histograms of the CG-score for various datasets to demonstrate that the CG-score can measure irregularity of the instances. We also demonstrate that the instances with higher CG-score are 'difficult' examples, which are learned slowly by the models, by comparing the loss and test accuracy curves and the evolution of Neural Tangent Kernel (NTK) submatrices of lowest/highest-scoring groups. | Published as a conference paper at ICLR 2023 DATA VALUATION WITHOUT TRAINING OF A MODEL |
d46935943 | Stochastic descent methods (of the gradient and mirror varieties) have become increasingly popular in optimization. In fact, it is now widely recognized that the success of deep learning is not only due to the special deep architecture of the models, but also due to the behavior of the stochastic descent methods used, which play a key role in reaching "good" solutions that generalize well to unseen data. In an attempt to shed some light on why this is the case, we revisit some minimax properties of stochastic gradient descent (SGD) for the square loss of linear models-originally developed in the 1990's-and extend them to general stochastic mirror descent (SMD) algorithms for general loss functions and nonlinear models. In particular, we show that there is a fundamental identity which holds for SMD (and SGD) under very general conditions, and which implies the minimax optimality of SMD (and SGD) for sufficiently small step size, and for a general class of loss functions and general nonlinear models. We further show that this identity can be used to naturally establish other properties of SMD (and SGD), namely convergence and implicit regularization for over-parameterized linear models (in what is now being called the "interpolating regime"), some of which have been shown in certain cases in prior literature. We also argue how this identity can be used in the so-called "highly over-parameterized" nonlinear setting (where the number of parameters far exceeds the number of data points) to provide insights into why SMD (and SGD) may have similar convergence and implicit regularization properties for deep learning. | STOCHASTIC GRADIENT/MIRROR DESCENT: MINI- MAX OPTIMALITY AND IMPLICIT REGULARIZATION |
d257050266 | Dynamic diagnosis is desirable when medical tests are costly or time-consuming. In this work, we use reinforcement learning (RL) to find a dynamic policy that selects lab test panels sequentially based on previous observations, ensuring accurate testing at a low cost. Clinical diagnostic data are often highly imbalanced; therefore, we aim to maximize the F1 score instead of the error rate. However, optimizing the non-concave F 1 score is not a classic RL problem, thus invalidates standard RL methods. To remedy this issue, we develop a reward shaping approach, leveraging properties of the F 1 score and duality of policy optimization, to provably find the set of all Pareto-optimal policies for budget-constrained F 1 score maximization. To handle the combinatorially complex state space, we propose a Semi-Model-based Deep Diagnosis Policy Optimization (SM-DDPO) framework that is compatible with end-to-end training and online learning. SM-DDPO is tested on diverse clinical tasks: ferritin abnormality detection, sepsis mortality prediction, and acute kidney injury diagnosis. Experiments with real-world data validate that SM-DDPO trains efficiently and identifies all Pareto-front solutions. Across all tasks, SM-DDPO is able to achieve state-of-the-art diagnosis accuracy (in some cases higher than conventional methods) with up to 85% reduction in testing cost. Core codes are available on GitHub 1 . | Published as a conference paper at ICLR 2023 DEEP REINFORCEMENT LEARNING FOR COST-EFFECTIVE MEDICAL DIAGNOSIS |
d244709251 | We propose Characteristic-Neural Ordinary Differential Equations (C-NODEs), a framework for extending Neural Ordinary Differential Equations (NODEs) beyond ODEs. While NODEs model the evolution of a latent variables as the solution to an ODE, C-NODE models the evolution of the latent variables as the solution of a family of first-order quasi-linear partial differential equations (PDEs) along curves on which the PDEs reduce to ODEs, referred to as characteristic curves. This in turn allows the application of the standard frameworks for solving ODEs, namely the adjoint method. Learning optimal characteristic curves for given tasks improves the performance and computational efficiency, compared to state of the art NODE models. We prove that the C-NODE framework extends the classical NODE on classification tasks by demonstrating explicit C-NODE representable functions not expressible by NODEs. Additionally, we present C-NODE-based continuous normalizing flows, which describe the density evolution of latent variables along multiple dimensions. Empirical results demonstrate the improvements provided by the proposed method for classification and density estimation on CIFAR-10, SVHN, and MNIST datasets under a similar computational budget as the existing NODE methods. The results also provide empirical evidence that the learned curves improve the efficiency of the system through a lower number of parameters and function evaluations compared with baselines. * | Characteristic Neural Ordinary Differential Equations |
d13268684 | There has been a lot of prior work on representation learning for speech recognition applications, but not much emphasis has been given to an investigation of effective representations of affect from speech, where the paralinguistic elements of speech are separated out from the verbal content. In this paper, we explore denoising autoencoders for learning paralinguistic attributes, i.e. categorical and dimensional affective traits from speech. We show that the representations learnt by the bottleneck layer of the autoencoder are highly discriminative of activation intensity and at separating out negative valence (sadness and anger) from positive valence (happiness). We experiment with different input speech features (such as FFT and log-mel spectrograms with temporal context windows), and different autoencoder architectures (such as stacked and deep autoencoders). We also learn utterance specific representations by a combination of denoising autoencoders and BLSTM based recurrent autoencoders. Emotion classification is performed with the learnt temporal/dynamic representations to evaluate the quality of the representations. Experiments on a well-established real-life speech dataset (IEMO-CAP) show that the learnt representations are comparable to state of the art feature extractors (such as voice quality features and MFCCs) and are competitive with state-of-the-art approaches at emotion and dimensional affect recognition. | Workshop track -ICLR 2016 LEARNING REPRESENTATIONS OF AFFECT FROM SPEECH |
d249209690 | The performance of deep neural networks is strongly influenced by the training dataset setup. In particular, when attributes with a strong correlation with the target attribute are present, the trained model can provide unintended prejudgments and show significant inference errors (i.e., the dataset bias problem). Various methods have been proposed to mitigate dataset bias, and their emphasis is on weakly correlated samples, called bias-conflicting samples. These methods are based on explicit bias labels provided by humans. However, such methods require human costs. Recently, several studies have sought to reduce human intervention by utilizing the output space values of neural networks, such as feature space, logits, loss, or accuracy. However, these output space values may be insufficient for the model to understand the bias attributes well. In this study, we propose a debiasing algorithm leveraging gradient called Per-sample Gradient-based Debiasing (PGD). PGD is comprised of three steps: (1) training a model on uniform batch sampling, (2) setting the importance of each sample in proportion to the norm of the sample gradient, and (3) training the model using importance-batch sampling, whose probability is obtained in step (2). Compared with existing baselines for various datasets, the proposed method showed state-of-the-art accuracy for the classification task. Furthermore, we describe theoretical understandings of how PGD can mitigate dataset bias. Code is available at Link * Equal contribution arXiv:2205.15704v3 [cs.LG] | Published as a conference paper at ICLR 2023 MITIGATING DATASET BIAS BY USING PER-SAMPLE GRADIENT |
d257365187 | The success of large-scale contrastive vision-language pretraining (CLIP) has benefited both visual recognition and multimodal content understanding. The concise design brings CLIP the advantage in inference efficiency against other visionlanguage models with heavier cross-attention fusion layers, making it a popular choice for a wide spectrum of downstream tasks. However, CLIP does not explicitly capture the hierarchical nature of high-level and fine-grained semantics conveyed in images and texts, which is arguably critical to vision-language understanding and reasoning. To this end, we equip both the visual and language branches in CLIP with hierarchy-aware attentions, namely Hierarchy-aware CLIP (HiCLIP), to progressively discover semantic hierarchies layer-by-layer from both images and texts in an unsupervised manner. As a result, such hierarchical aggregation significantly improves the cross-modal alignment. To demonstrate the advantages of HiCLIP, we conduct qualitative analysis on its unsupervised hierarchy induction during inference, as well as extensive quantitative experiments on both visual recognition and vision-language downstream tasks. 1 * This work was conducted while interning at ByteDance. 1 We release our implementation of HiCLIP at https://github.com/jeykigung/HiCLIP. | Published as a conference paper at ICLR 2023 HICLIP: CONTRASTIVE LANGUAGE-IMAGE PRE- TRAINING WITH HIERARCHY-AWARE ATTENTION |
d250072832 | Content creators compete for user attention. Their reach crucially depends on algorithmic choices made by developers on online platforms. To maximize exposure, many creators adapt strategically, as evidenced by examples like the sprawling search engine optimization industry. This begets competition for the finite user attention pool. We formalize these dynamics in what we call an exposure game, a model of incentives induced by algorithms, including modern factorization and (deep) two-tower architectures. We prove that seemingly innocuous algorithmic choices-e.g., non-negative vs. unconstrained factorization-significantly affect the existence and character of (Nash) equilibria in exposure games. We proffer use of creator behavior models, like exposure games, for an (ex-ante) predeployment audit. Such an audit can identify misalignment between desirable and incentivized content, and thus complement post-hoc measures like content filtering and moderation. To this end, we propose tools for numerically finding equilibria in exposure games, and illustrate results of an audit on the MovieLens and LastFM datasets. Among else, we find that the strategically produced content exhibits strong dependence between algorithmic exploration and content diversity, and between model expressivity and bias towards gender-based user and creator groups. | Published as a conference paper at ICLR 2023 MODELING CONTENT CREATOR INCENTIVES ON ALGORITHM-CURATED PLATFORMS |
d195820512 | Adversarial examples raise questions about whether neural network models are sensitive to the same visual features as humans. In this paper, we first detect adversarial examples or otherwise corrupted images based on a class-conditional reconstruction of the input. To specifically attack our detection mechanism, we propose the Reconstructive Attack which seeks both to cause a misclassification and a low reconstruction error. This reconstructive attack produces undetected adversarial examples but with much smaller success rate. Among all these attacks, we find that CapsNets always perform better than convolutional networks. Then, we diagnose the adversarial examples for CapsNets and find that the success of the reconstructive attack is highly related to the visual similarity between the source and target class. Additionally, the resulting perturbations can cause the input image to appear visually more like the target class and hence become non-adversarial. This suggests that CapsNets use features that are more aligned with human perception and have the potential to address the central issue raised by adversarial examples. * Equal Contributions. | Published as a conference paper at ICLR 2020 DETECTING AND DIAGNOSING ADVERSARIAL IMAGES WITH CLASS-CONDITIONAL CAPSULE RECONSTRUCTIONS |
d257232577 | We address the challenging problem of jointly inferring the 3D flow and volumetric densities moving in a fluid from a monocular input video with a deep neural network. Despite the complexity of this task, we show that it is possible to train the corresponding networks without requiring any 3D ground truth for training. In the absence of ground truth data we can train our model with observations from realworld capture setups instead of relying on synthetic reconstructions. We make this unsupervised training approach possible by first generating an initial prototype volume which is then moved and transported over time without the need for volumetric supervision. Our approach relies purely on image-based losses, an adversarial discriminator network, and regularization. Our method can estimate long-term sequences in a stable manner, while achieving closely matching targets for inputs such as rising smoke plumes. | Published as a conference paper at ICLR 2023 LEARNING TO ESTIMATE SINGLE-VIEW VOLUMETRIC FLOW MOTIONS WITHOUT 3D SUPERVISION |
d235613377 | Despite the success of recent Neural Architecture Search (NAS) methods on various tasks which have shown to output networks that largely outperform humandesigned networks, conventional NAS methods have mostly tackled the optimization of searching for the network architecture for a single task (dataset), which does not generalize well across multiple tasks (datasets). Moreover, since such task-specific methods search for a neural architecture from scratch for every given task, they incur a large computational cost, which is problematic when the time and monetary budget are limited. In this paper, we propose an efficient NAS framework that is trained once on a database consisting of datasets and pretrained networks and can rapidly search for a neural architecture for a novel dataset. The proposed MetaD2A (Meta Dataset-to-Architecture) model can stochastically generate graphs (architectures) from a given set (dataset) via a cross-modal latent space learned with amortized meta-learning. Moreover, we also propose a meta-performance predictor to estimate and select the best architecture without direct training on target datasets. The experimental results demonstrate that our model meta-learned on subsets of ImageNet-1K and architectures from NAS-Bench 201 search space successfully generalizes to multiple unseen datasets including CIFAR-10 and CIFAR-100, with an average search time of 33 GPU seconds. Even under MobileNetV3 search space, MetaD2A is 5.5K times faster than NSGANetV2, a transferable NAS method, with comparable performance. We believe that the MetaD2A proposes a new research direction for rapid NAS as well as ways to utilize the knowledge from rich databases of datasets and architectures accumulated over the past years. Code is available at https://github.com/HayeonLee/MetaD2A. * These authors contributed equally to this work. arXiv:2107.00860v1 [cs.LG] 2 Jul 2021 Published as a conference paper at ICLR 2021 Conventional NAS Approach Training NAS Model NAS Model NAS Model | Published as a conference paper at ICLR 2021 RAPID NEURAL ARCHITECTURE SEARCH BY LEARNING TO GENERATE GRAPHS FROM DATASETS |
d235368380 | The empirical success of deep convolutional networks on tasks involving highdimensional data such as images or audio suggests that they can efficiently approximate certain functions that are well-suited for such tasks. In this paper, we study this through the lens of kernel methods, by considering simple hierarchical kernels with two or three convolution and pooling layers, inspired by convolutional kernel networks. These achieve good empirical performance on standard vision datasets, while providing a precise description of their functional space that yields new insights on their inductive bias. We show that the RKHS consists of additive models of interaction terms between patches, and that its norm encourages spatial similarities between these terms through pooling layers. We then provide generalization bounds which illustrate how pooling and patches yield improved sample complexity guarantees when the target function presents such regularities. | Published as a conference paper at ICLR 2022 APPROXIMATION AND LEARNING WITH DEEP CONVOLUTIONAL MODELS: A KERNEL PERSPECTIVE |
d247011143 | Recent studies demonstrate that deep networks, even robustified by the state-ofthe-art adversarial training (AT), still suffer from large robust generalization gaps, in addition to the much more expensive training costs than standard training. In this paper, we investigate this intriguing problem from a new perspective, i.e., injecting appropriate forms of sparsity during adversarial training. We introduce two alternatives for sparse adversarial training: (i) static sparsity, by leveraging recent results from the lottery ticket hypothesis to identify critical sparse subnetworks arising from the early training; (ii) dynamic sparsity, by allowing the sparse subnetwork to adaptively adjust its connectivity pattern (while sticking to the same sparsity ratio) throughout training. We find both static and dynamic sparse methods to yield win-win: substantially shrinking the robust generalization gap and alleviating the robust overfitting, meanwhile significantly saving training and inference FLOPs. Extensive experiments validate our proposals with multiple network architectures on diverse datasets, including CIFAR-10/100 and Tiny-ImageNet. For example, our methods reduce robust generalization gap and overfitting by 34.44% and 4.02%, with comparable robust/standard accuracy boosts and 87.83%/87.82% training/inference FLOPs savings on CIFAR-100 with ResNet-18. Besides, our approaches can be organically combined with existing regularizers, establishing new state-of-the-art results in AT. Codes are available in https: //github.com/VITA-Group/Sparsity-Win-Robust-Generalization. | Published as a conference paper at ICLR 2022 SPARSITY WINNING TWICE: BETTER ROBUST GEN- ERALIZATION FROM MORE EFFICIENT TRAINING |
d6530726 | We study the problem of building models that disentangle independent factors of variation. Such models could be used to encode features that can efficiently be used for classification and to transfer attributes between different images in image synthesis. As data we use a weakly labeled training set. Our weak labels indicate what single factor has changed between two data samples, although the relative value of the change is unknown. This labeling is of particular interest as it may be readily available without annotation costs. To make use of weak labels we introduce an autoencoder model and train it through constraints on image pairs and triplets. We formally prove that without additional knowledge there is no guarantee that two images with the same factor of variation will be mapped to the same feature. We call this issue the reference ambiguity. Moreover, we show the role of the feature dimensionality and adversarial training. We demonstrate experimentally that the proposed model can successfully transfer attributes on several datasets, but show also cases when the reference ambiguity occurs. * The authors contributed equally. | CHALLENGES IN DISENTANGLING INDEPENDENT FAC- TORS OF VARIATION * |
d251648016 | Spiking Neural Networks (SNNs) have gained great attraction due to their distinctive properties of low power consumption and fast inference on neuromorphic hardware. As the most effective method to get deep SNNs, ANN-SNN conversion has achieved comparable performance as ANNs on large-scale datasets. Despite this, it requires long time-steps to match the firing rates of SNNs to the activation of ANNs. As a result, the converted SNN suffers severe performance degradation problems with short time-steps, which hamper the practical application of SNNs. In this paper, we theoretically analyze ANN-SNN conversion error and derive the estimated activation function of SNNs. Then we propose the quantization clipfloor-shift activation function to replace the ReLU activation function in source ANNs, which can better approximate the activation function of SNNs. We prove that the expected conversion error between SNNs and ANNs is zero, enabling us to achieve high-accuracy and ultra-low-latency SNNs. We evaluate our method on CIFAR-10/100 and ImageNet datasets, and show that it outperforms the stateof-the-art ANN-SNN and directly trained SNNs in both accuracy and time-steps. To the best of our knowledge, this is the first time to explore high-performance ANN-SNN conversion with ultra-low latency (4 time-steps). Code is available at https://github.com/putshua/SNN conversion QCFS Published as a conference paper at ICLR 2022 | OPTIMAL ANN-SNN CONVERSION FOR HIGH- ACCURACY AND ULTRA-LOW-LATENCY SPIKING NEURAL NETWORKS |
d8208459 | Reinforcement learning is concerned with identifying reward-maximizing behaviour policies in environments that are initially unknown. Stateof-the-art reinforcement learning approaches, such as deep Q-networks, are model-free and learn to act effectively across a wide range of environments such as Atari games, but require huge amounts of data. Model-based techniques are more data-efficient, but need to acquire explicit knowledge about the environment.In this paper, we take a step towards using modelbased techniques in environments with a highdimensional visual state space by demonstrating that it is possible to learn system dynamics and the reward structure jointly. Our contribution is to extend a recently developed deep neural network for video frame prediction in Atari games to enable reward prediction as well. To this end, we phrase a joint optimization problem for minimizing both video frame and reward reconstruction loss, and adapt network parameters accordingly. Empirical evaluations on five Atari games demonstrate accurate cumulative reward prediction of up to 200 frames. We consider these results as opening up important directions for model-based reinforcement learning in complex, initially unknown environments. | A Deep Learning Approach for Joint Video Frame and Reward Prediction in Atari Games |
d53015027 | Deep learning has shown high performances in various types of tasks from visual recognition to natural language processing, which indicates superior flexibility and adaptivity of deep learning. To understand this phenomenon theoretically, we develop a new approximation and estimation error analysis of deep learning with the ReLU activation for functions in a Besov space and its variant with mixed smoothness. The Besov space is a considerably general function space including the Hölder space and Sobolev space, and especially can capture spatial inhomogeneity of smoothness. Through the analysis in the Besov space, it is shown that deep learning can achieve the minimax optimal rate and outperform any nonadaptive (linear) estimator such as kernel ridge regression, which shows that deep learning has higher adaptivity to the spatial inhomogeneity of the target function than other estimators such as linear ones. In addition to this, it is shown that deep learning can avoid the curse of dimensionality if the target function is in a mixed smooth Besov space. We also show that the dependency of the convergence rate on the dimensionality is tight due to its minimax optimality. These results support high adaptivity of deep learning and its superior ability as a feature extractor. | ADAPTIVITY OF DEEP RELU NETWORK FOR LEARN- ING IN BESOV AND MIXED SMOOTH BESOV SPACES: OPTIMAL RATE AND CURSE OF DIMENSIONALITY |
d256459906 | Diffusion models generate samples by reversing a fixed forward diffusion process. Despite already providing impressive empirical results, these diffusion models algorithms can be further improved by reducing the variance of the training targets in their denoising score-matching objective. We argue that the source of such variance lies in the handling of intermediate noise-variance scales, where multiple modes in the data affect the direction of reverse paths. We propose to remedy the problem by incorporating a reference batch which we use to calculate weighted conditional scores as more stable training targets. We show that the procedure indeed helps in the challenging intermediate regime by reducing (the trace of) the covariance of training targets. The new stable targets can be seen as trading bias for reduced variance, where the bias vanishes with increasing reference batch size. Empirically, we show that the new objective improves the image quality, stability, and training speed of various popular diffusion models across datasets with both general ODE and SDE solvers. When used in combination with EDM (Karras et al., 2022), our method yields a current SOTA FID of 1.90 with 35 network evaluations on the unconditional CIFAR-10 generation task. The code is available at https://github.com/Newbeeer/stf * Equal Contribution. | Published as a conference paper at ICLR 2023 STABLE TARGET FIELD FOR REDUCED VARIANCE SCORE ESTIMATION IN DIFFUSION MODELS |
d257766694 | Dense prediction tasks are a fundamental class of problems in computer vision. As supervised methods suffer from high pixel-wise labeling cost, a few-shot learning solution that can learn any dense task from a few labeled images is desired. Yet, current few-shot learning methods target a restricted set of tasks such as semantic segmentation, presumably due to challenges in designing a general and unified model that is able to flexibly and efficiently adapt to arbitrary tasks of unseen semantics. We propose Visual Token Matching (VTM), a universal few-shot learner for arbitrary dense prediction tasks. It employs non-parametric matching on patchlevel embedded tokens of images and labels that encapsulates all tasks. Also, VTM flexibly adapts to any task with a tiny amount of task-specific parameters that modulate the matching algorithm. We implement VTM as a powerful hierarchical encoder-decoder architecture involving ViT backbones where token matching is performed at multiple feature hierarchies. We experiment VTM on a challenging variant of Taskonomy dataset and observe that it robustly few-shot learns various unseen dense prediction tasks. Surprisingly, it is competitive with fully supervised baselines using only 10 labeled examples of novel tasks (0.004% of full supervision) and sometimes outperforms using 0.1% of full supervision. Codes are available at https://github.com/GitGyun/visual_token_matching. | UNIVERSAL FEW-SHOT LEARNING OF DENSE PREDIC- TION TASKS WITH VISUAL TOKEN MATCHING |
d212725580 | Adversarial patch attacks are among of the most practical threat models against realworld computer vision systems. This paper studies certified and empirical defenses against patch attacks. We begin with a set of experiments showing that most existing defenses, which work by pre-processing input images to mitigate adversarial patches, are easily broken by simple white-box adversaries. Motivated by this finding, we propose the first certified defense against patch attacks, and propose faster methods for its training. Furthermore, we experiment with different patch shapes for testing, obtaining surprisingly good robustness transfer across shapes, and present preliminary results on certified defense against sparse attacks. Our complete implementation can be found on: https://github.com/Ping-C/ certifiedpatchdefense. | Published as a conference paper at ICLR 2020 CERTIFIED DEFENSES FOR ADVERSARIAL PATCHES * |
d203591519 | Despite alarm over the reliance of machine learning systems on so-called spurious patterns, the term lacks coherent meaning in standard statistical frameworks. However, the language of causality offers clarity: spurious associations are due to confounding (e.g., a common cause), but not direct or indirect causal effects. In this paper, we focus on natural language processing, introducing methods and resources for training models less sensitive to spurious patterns. Given documents and their initial labels, we task humans with revising each document so that it (i) accords with a counterfactual target label; (ii) retains internal coherence; and (iii) avoids unnecessary changes. Interestingly, on sentiment analysis and natural language inference tasks, classifiers trained on original data fail on their counterfactually-revised counterparts and vice versa. Classifiers trained on combined datasets perform remarkably well, just shy of those specialized to either domain. While classifiers trained on either original or manipulated data alone are sensitive to spurious features (e.g., mentions of genre), models trained on the combined data are less sensitive to this signal. Both datasets are publicly available 1 . | Published as a conference paper at ICLR 2020 LEARNING THE DIFFERENCE THAT MAKES A DIFFER- ENCE WITH COUNTERFACTUALLY-AUGMENTED DATA |
d252683397 | Reward and representation learning are two long-standing challenges for learning an expanding set of robot manipulation skills from sensory observations. Given the inherent cost and scarcity of in-domain, task-specific robot data, learning from large, diverse, offline human videos has emerged as a promising path towards acquiring a generally useful visual representation for control; however, how these human videos can be used for general-purpose reward learning remains an open question. We introduce Value-Implicit Pre-training (VIP), a self-supervised pre-trained visual representation capable of generating dense and smooth reward functions for unseen robotic tasks. VIP casts representation learning from human videos as an offline goal-conditioned reinforcement learning problem and derives a selfsupervised goal-conditioned value-function objective that does not depend on actions, enabling pre-training on unlabeled human videos. Theoretically, VIP can be understood as a novel implicit time contrastive objective that generates a temporally smooth embedding, enabling the value function to be implicitly defined via the embedding distance, which can then be used to construct the reward function for any goal-image specified downstream task. Trained on large-scale Ego4D human videos and without any fine-tuning on in-domain, task-specific data, VIP can provide dense visual reward for an extensive set of simulated and real-robot tasks, enabling diverse reward-based visual control methods and outperforming all prior pre-trained representations. Notably, VIP can enable simple, few-shot offline RL on a suite of real-world robot tasks with as few as 20 trajectories. . Persim: Data-efficient offline reinforcement learning with heterogeneous agents via personalized simulators. | Published as a conference paper at ICLR 2023 TOWARDS UNIVERSAL VISUAL REWARD AND REPRE- SENTATION VIA VALUE-IMPLICIT PRE-TRAINING |
d14079971 | Recent studies have shown that deep neural networks (DNN) are vulnerable to adversarial samples: maliciously-perturbed samples crafted to yield incorrect model outputs. Such attacks can severely undermine DNN systems, particularly in security-sensitive settings. It was observed that an adversary could easily generate adversarial samples by making a small perturbation on irrelevant feature dimensions that are unnecessary for the current classification task. To overcome this problem, we introduce a defensive mechanism called DeepCloak. By identifying and removing unnecessary features in a DNN model, DeepCloak limits the capacity an attacker can use generating adversarial samples and therefore increase the robustness against such inputs. Comparing with other defensive approaches, DeepCloak is easy to implement and computationally efficient. Experimental results show that DeepCloak can increase the performance of DNN models against adversarial samples. | Workshop track -ICLR 2017 DEEPCLOAK: MASKING DNN MODELS FOR ROBUST- NESS AGAINST ADVERSARIAL SAMPLES |
d19326248 | Genetic algorithms have been widely used in many practical optimization problems. Inspired by natural selection, operators, including mutation, crossover and selection, provide effective heuristics for search and black-box optimization. However, they have not been shown useful for deep reinforcement learning, possibly due to the catastrophic consequence of parameter crossovers of neural networks. Here, we present Genetic Policy Optimization (GPO), a new genetic algorithm for sample-efficient deep policy optimization. GPO uses imitation learning for policy crossover in the state space and applies policy gradient methods for mutation. Our experiments on Mujoco tasks show that GPO as a genetic algorithm is able to provide superior performance over the state-of-the-art policy gradient methods and achieves comparable or higher sample efficiency. | GENETIC POLICY OPTIMIZATION |
d254591489 | We present RAVEn, a self-supervised multi-modal approach to jointly learn visual and auditory speech representations. Our pre-training objective involves encoding masked inputs, and then predicting contextualised targets generated by slowly-evolving momentum encoders. Driven by the inherent differences between video and audio, our design is asymmetric w.r.t. the two modalities' pretext tasks: Whereas the auditory stream predicts both the visual and auditory targets, the visual one predicts only the auditory targets. We observe strong results in low-and high-resource labelled data settings when fine-tuning the visual and auditory encoders resulting from a single pre-training stage, in which the encoders are jointly trained. Notably, RAVEn surpasses all self-supervised methods on visual speech recognition (VSR) on LRS3, and combining RAVEn with self-training using only 30 hours of labelled data even outperforms a recent semi-supervised method trained on 90,000 hours of non-public data. At the same time, we achieve state-of-the-art results in the LRS3 low-resource setting for auditory speech recognition (as well as for VSR). Our findings point to the viability of learning powerful speech representations entirely from raw video and audio, i.e., without relying on handcrafted features. Code and models are available at | Published as a conference paper at ICLR 2023 JOINTLY LEARNING VISUAL AND AUDITORY SPEECH REPRESENTATIONS FROM RAW DATA |
d244117525 | Unsupervised large-scale vision-language pre-training has shown promising advances on various downstream tasks. Existing methods often model the crossmodal interaction either via the similarity of the global feature of each modality which misses sufficient information, or finer-grained interactions using cross/selfattention upon visual and textual tokens. However, cross/self-attention suffers from inferior efficiency in both training and inference. In this paper, we introduce a large-scale Fine-grained Interactive Language-Image Pre-training (FILIP) to achieve finer-level alignment through a cross-modal late interaction mechanism, which uses a token-wise maximum similarity between visual and textual tokens to guide the contrastive objective. FILIP successfully leverages the finergrained expressiveness between image patches and textual words by modifying only contrastive loss, while simultaneously gaining the ability to pre-compute image and text representations offline at inference, keeping both large-scale training and inference efficient. Furthermore, we construct a new large-scale image-text pair dataset called FILIP300M for pre-training. Experiments show that FILIP achieves state-of-the-art performance on multiple downstream vision-language tasks including zero-shot image classification and image-text retrieval. The visualization on word-patch alignment further shows that FILIP can learn meaningful fine-grained features with promising localization ability. * Equal contribution † Corresponding authors: xu.hang@huawei.com, | FILIP: FINE-GRAINED INTERACTIVE LANGUAGE- IMAGE PRE-TRAINING |
d13747555 | Deep residual networks (ResNets) made a recent breakthrough in deep learning. The core idea of ResNets is to have shortcut connections between layers that allow the network to be much deeper while still being easy to optimize avoiding vanishing gradients. These shortcut connections have interesting side-effects that make ResNets behave differently from other typical network architectures. In this work we use these properties to design a network based on a ResNet but with parameter sharing and with adaptive computation time. The resulting network is much smaller than the original network and can adapt the computational cost to the complexity of the input image. * The research work was done during an internship at NVIDIA research, | Workshop track -ICLR 2018 IAMNN: ITERATIVE AND ADAPTIVE MOBILE NEURAL NETWORK FOR EFFICIENT IMAGE CLASSIFICATION |
d248524779 | Differentiable physics has recently been shown as a powerful tool for solving soft-body manipulation tasks. However, the differentiable physics solver often gets stuck when the initial contact points of the end effectors are sub-optimal or when performing multi-stage tasks that require contact point switching, which often leads to local minima. To address this challenge, we propose a contact point discovery approach (CPDeform) that guides the stand-alone differentiable physics solver to deform various soft-body plasticines. The key idea of our approach is to integrate optimal transport-based contact points discovery into the differentiable physics solver to overcome the local minima from initial contact points or contact switching. On single-stage tasks, our method can automatically find suitable initial contact points based on transport priorities. On complex multi-stage tasks, we can iteratively switch the contact points of end-effectors based on transport priorities. To evaluate the effectiveness of our method, we introduce PlasticineLab-M that extends the existing differentiable physics benchmark PlasticineLab to seven new challenging multi-stage soft-body manipulation tasks. Extensive experimental results suggest that: 1) on multi-stage tasks that are infeasible for the vanilla differentiable physics solver, our approach discovers contact points that efficiently guide the solver to completion; 2) on tasks where the vanilla solver performs sub-optimally or near-optimally, our contact point discovery method performs better than or on par with the manipulation performance | Published as a conference paper at ICLR 2022 CONTACT POINTS DISCOVERY FOR SOFT-BODY MA- NIPULATIONS WITH DIFFERENTIABLE PHYSICS |
d24044364 | Recently, the problem of local minima in very high dimensional non-convex optimization has been challenged and the problem of saddle points has been introduced. This paper introduces a dynamic type of normalization that forces the system to escape saddle points. Unlike other saddle point escaping algorithms, second order information is not utilized, and the system can be trained with an arbitrary gradient descent learner. The system drastically improves learning in a range of deep neural networks on various data-sets in comparison to non-CPN neural networks. | Workshop track -ICLR 2017 CHARGED POINT NORMALIZATION AN EFFICIENT SOLUTION TO THE SADDLE POINT PROBLEM |
d247839322 | We present a meta-learning framework for learning new visual concepts quickly, from just one or a few examples, guided by multiple naturally occurring data streams: simultaneously looking at images, reading sentences that describe the objects in the scene, and interpreting supplemental sentences that relate the novel concept with other concepts. The learned concepts support downstream applications, such as answering questions by reasoning about unseen images. Our model, namely FALCON, represents individual visual concepts, such as colors and shapes, as axis-aligned boxes in a high-dimensional space (the "box embedding space"). Given an input image and its paired sentence, our model first resolves the referential expression in the sentence and associates the novel concept with particular objects in the scene. Next, our model interprets supplemental sentences to relate the novel concept with other known concepts, such as "X has property Y" or "X is a kind of Y". Finally, it infers an optimal box embedding for the novel concept that jointly 1) maximizes the likelihood of the observed instances in the image, and 2) satisfies the relationships between the novel concepts and the known ones. We demonstrate the effectiveness of our model on both synthetic and real-world datasets. * Equal contribution. | Published as a conference paper at ICLR 2022 FALCON: FAST VISUAL CONCEPT LEARNING BY IN- TEGRATING IMAGES, LINGUISTIC DESCRIPTIONS, AND CONCEPTUAL RELATIONS |
d2753399 | For most deep learning algorithms training is notoriously time consuming. Since most of the computation in training neural networks is typically spent on floating point multiplications, we investigate an approach to training that eliminates the need for most of these. Our method consists of two parts: First we stochastically binarize weights to convert multiplications involved in computing hidden states to sign changes. Second, while back-propagating error derivatives, in addition to binarizing the weights, we quantize the representations at each layer to convert the remaining multiplications into binary shifts. Experimental results across 3 popular datasets (MNIST, CIFAR10, SVHN) show that this approach not only does not hurt classification performance but can result in even better performance than standard stochastic gradient descent training, paving the way to fast, hardwarefriendly training of neural networks. | Published as a conference paper at ICLR 2016 NEURAL NETWORKS WITH FEW MULTIPLICATIONS |
d257771315 | Recent works have shown that sequence modeling can be effectively used to train reinforcement learning (RL) policies. However, the success of applying existing sequence models to planning, in which we wish to obtain a trajectory of actions to reach some goal, is less straightforward. The typical autoregressive generation procedures of sequence models preclude sequential refinement of earlier steps, which limits the effectiveness of a predicted plan. In this paper, we suggest an approach towards integrating planning with sequence models based on the idea of iterative energy minimization, and illustrate how such a procedure leads to improved RL performance across different tasks. We train a masked language model to capture an implicit energy function over trajectories of actions, and formulate planning as finding a trajectory of actions with minimum energy. We illustrate how this procedure enables improved performance over recent approaches across BabyAI and Atari environments. We further demonstrate unique benefits of our iterative optimization procedure, involving new task generalization, test-time constraints adaptation, and the ability to compose plans together. Project website: https://hychen-naza.github.io/projects/LEAP.Published as a conference paper at ICLR 2023 timization procedure on an energy function over trajectories defined implicitly by a sequence model (illustrated inFigure 1). To define an energy function across trajectories, we train a bidirectional sequence model using a masked-language modeling (MLM) objective (Devlin et al., 2019). We define the energy of a trajectory as the negative pseudo-likelihood (PLL) of this MLM (Salazar et al., 2019) and sequentially minimize this energy value by replacing actions at different timepoints in the trajectory with the marginal estimates given by the MLM. Since our MLM is bi-directional in nature, the choice of new action at a given time-step is generated based on both future and past actions.By iteratively generating actions through planning, we illustrate how our proposed framework outperforms prior methods in both BabyAI(Chevalier-Boisvert et al., 2019)and Atari (Bellemare et al., 2013) tasks. Furthermore, by formulating the action generation process as an iterative energy minimization procedure, we illustrate how this enables us to generalize to environments with new sets of test-time constraints as well as more complex planning problems. Finally, we demonstrate how such an energy minimization procedure enables us to compose planning procedures in different models together, enabling the construction of plan which achieves multiple objectives.Concretely, in this paper, we contribute the following: First, we introduce LEAP, a framework through which we may iteratively plan with sequence models. Second, we illustrate how such a planning framework can be beneficial on both BabyAI and Atari domains. Finally, we illustrate how iteratively planning through energy minimization gives a set of unique properties, enabling better test time performance on more complex environments and environments with new test-time obstacles, and the ability to compose multiple learned models together, to jointly generate plans that satisfy multiple sets of goals. | Published as a conference paper at ICLR 2023 PLANNING WITH SEQUENCE MODELS THROUGH ITER- ATIVE ENERGY MINIMIZATION |
d251280143 | In this paper, we study how to use masked signal modeling in vision and language (V+L) representation learning. Instead of developing masked language modeling (MLM) and masked image modeling (MIM) independently, we propose to build joint masked vision and language modeling, where the masked signal of one modality is reconstructed with the help from another modality. This is motivated by the nature of image-text paired data that both of the image and the text convey almost the same information but in different formats. The masked signal reconstruction of one modality conditioned on another modality can also implicitly learn cross-modal alignment between language tokens and image patches. Our experiments on various V+L tasks show that the proposed method, along with common V+L alignment losses, achieves state-of-the-art performance in the regime of millions of pre-training data. Also, we outperforms the other competitors by a significant margin in limited data scenarios. | Published as a conference paper at ICLR 2023 MASKED VISION AND LANGUAGE MODELING FOR MULTI-MODAL REPRESENTATION LEARNING |
d231740588 | Knowledge distillation is an effective approach to leverage a well-trained network or an ensemble of them, named as the teacher, to guide the training of a student network. The outputs from the teacher network are used as soft labels for supervising the training of a new network. Recent studies(Müller et al., 2019;Yuan et al., 2020)revealed an intriguing property of the soft labels that making labels soft serves as a good regularization to the student network. From the perspective of statistical learning, regularization aims to reduce the variance, however how bias and variance change is not clear for training with soft labels. In this paper, we investigate the bias-variance tradeoff brought by distillation with soft labels. Specifically, we observe that during training the bias-variance tradeoff varies sample-wisely. Further, under the same distillation temperature setting, we observe that the distillation performance is negatively associated with the number of some specific samples, which are named as regularization samples since these samples lead to bias increasing and variance decreasing. Nevertheless, we empirically find that completely filtering out regularization samples also deteriorates distillation performance. Our discoveries inspired us to propose the novel weighted soft labels to help the network adaptively handle the sample-wise biasvariance tradeoff. Experiments on standard evaluation benchmarks validate the effectiveness of our method. Our code is available at https://github.com/ bellymonster/Weighted-Soft-Label-Distillation.Published as a conference paper at ICLR 2021 design a teacher-free distillation method by turning the predictions of the student network into soft labels. | Published as a conference paper at ICLR 2021 RETHINKING SOFT LABELS FOR KNOWLEDGE DISTIL- LATION: A BIAS-VARIANCE TRADEOFF PERSPECTIVE |
d2974620 | We present a representation learning method that learns features at multiple different levels of scale. Working within the unsupervised framework of denoising autoencoders, we observe that when the input is heavily corrupted during training, the network tends to learn coarse-grained features, whereas when the input is only slightly corrupted during training, the network tends to learn fine-grained features. This motivates the scheduled denoising autoencoder, which starts with a high level of input noise that lowers as training progresses. We find that the resulting representation yields a significant boost on a later supervised task compared to the original input, or to a standard denoising autoencoder trained at a single noise level. | Scheduled denoising autoencoders |
d232147821 | We propose a simple, general and effective technique, Reward Randomization for discovering diverse strategic policies in complex multi-agent games. Combining reward randomization and policy gradient, we derive a new algorithm, Reward-Randomized Policy Gradient (RPG). RPG is able to discover multiple distinctive human-interpretable strategies in challenging temporal trust dilemmas, including grid-world games and a real-world game Agar.io, where multiple equilibria exist but standard multi-agent policy gradient algorithms always converge to a fixed one with a sub-optimal payoff for every player even using state-of-the-art exploration techniques. Furthermore, with the set of diverse strategies from RPG, we can (1) achieve higher payoffs by fine-tuning the best policy from the set; and (2) obtain an adaptive agent by using this set of strategies as its training opponents. The source code and example videos can be found in our website: https://sites.google.Published as a conference paper at ICLR 2021 for both agents and a "safe" non-cooperative equilibrium with strictly lower payoffs. We show, from both theoretical and practical perspectives, that even in this simple matrix-form game, PG fails to discover the high-payoff "risky" NE with high probability. The intuition is that the neighborhood that makes policies converge to the "risky" NE can be substantially small comparing to the entire policy space. Therefore, an exponentially large number of exploration steps are needed to ensure PG discovers the desired mode. We propose a simple technique, Reward Randomization (RR), pay-off policy space risky coop. non-coop. Stag-Hunt pay-off coop. non-coop. Perturbed Game reward randomization evaluate in the orignal game policy space Figure 1: Intuition of Reward Randomizationwhich can help PG discover the "risky" cooperation strategy in the stag-hunt game with theoretical guarantees. The core idea of RR is to directly perturb the reward structure of the multi-agent game of interest, which is typically low-dimensional. RR directly alters the landscape of different strategy modes in the policy space and therefore makes it possible to easily discover novel behavior in the perturbed game (Fig. 1). We call this new PG variant Reward-Randomized Policy Gradient (RPG).To further illustrate the effectiveness of RPG, we introduce three Markov games -two gridworld games and a real-world online game Agar.io. All these games have multiple NEs including both "risky" cooperation strategies and "safe" non-cooperative strategies. We empirically show that even with state-of-the-art exploration techniques, PG fails to discover the "risky" cooperation strategies. In contrast, RPG discovers a surprisingly diverse set of human-interpretable strategies in all these games, including some non-trivial emergent behavior. Importantly, among this set are policies achieving much higher payoffs for each player compared to those found by PG. This "diversityseeking" property of RPG also makes it feasible to build adaptive policies: by re-training an RL agent against the diverse opponents discovered by RPG, the agent is able to dynamically alter its strategy between different modes, e.g., either cooperate or compete, w.r.t. its test-time opponent's behavior.We summarize our contributions as follow • We studied a collection of challenging multi-agent games, where the popular multi-agent PG algorithm always converges to a sub-optimal equilibrium strategy with low payoffs.• A novel reward-space exploration technique, reward randomization (RR), for discovering hard-to-find equilibrium with high payoffs. Both theoretical and empirical results show that reward randomization substantially outperforms classical policy/action-space exploration techniques in challenging trust dilemmas.• We empirically show that RR discovers surprisingly diverse strategic behaviors in complex Markov games, which further provides a practical solution for building an adaptive agent.• A new multi-agent environment Agar.io, which allows complex multi-agent strategic behavior. We released the environment to the community as a novel testbed for MARL research.A MOTIVATING EXAMPLE: STAG HUNTStag Hare Stag a, a c, b Hare b, c d, d Table 1: The stag-hunt game, a > b ≥ d > c.We start by analyzing a simple problem: finding the NE with the optimal payoffs in the Stag Hunt game. This game was originally introduced in Rousseau's work, "A discourse on inequality" (Rousseau, 1984): a group of hunters are tracking a big stag silently; now a hare shows up, each hunter should decide whether to keep tracking the stag or kill the hare immediately. This leads to the 2-by-2 matrix-form stag-hunt game in Tab. 1 with two actions for each agent, Stag (S) and Hare (H). There are two pure strategy NEs: the Stag NE, where both agents choose S and receive a high payoff a (e.g., a = 4), and the Hare NE, where both agents choose H and receive a lower payoff d (e.g., d = 1). The Stag NE is "risky" because if one agent defects, they still receives a decent reward b (e.g., b = 3) for eating the hare alone while the other agent with an S action may suffer from a big loss c for being hungry (e.g., c = −10).Formally, let A = {S, H} denote the action space, π i (θ i ) denote the policy for agent i (i ∈ {1, 2}) parameterized by θ i , i.e., P [π i (θ i ) = S] = θ i and P [π i (θ i ) = H] = 1 − θ i , and R(a 1 , a 2 ; i) denote the payoff for agent i when agent 1 takes action a 1 and agent 2 takes action a 2 . Each agent i optimizes its expected utility U i (π 1 , π 2 ) = E a1∼π1,a2∼π2 [R(a 1 , a 2 ; i)]. Using the standard policy gradient algorithm, a typical learning procedure is to repeatedly take the following two steps until | Published as a conference paper at ICLR 2021 DISCOVERING DIVERSE MULTI-AGENT STRATEGIC BEHAVIOR VIA REWARD RANDOMIZATION |
d26945723 | Neural networks have recently had a lot of success for many tasks. However, neural network architectures that perform well are still typically designed manually by experts in a cumbersome trial-and-error process. We propose a new method to automatically search for well-performing CNN architectures based on a simple hill climbing procedure whose operators apply network morphisms, followed by short optimization runs by cosine annealing. Surprisingly, this simple method yields competitive results, despite only requiring resources in the same order of magnitude as training a single network. E.g., on CIFAR-10, our method designs and trains networks with an error rate below 6% in only 12 hours on a single GPU; training for one day reduces this error further, to almost 5%. | Under review as a conference paper at ICLR 2018 SIMPLE AND EFFICIENT ARCHITECTURE SEARCH FOR CONVOLUTIONAL NEURAL NETWORKS |
d210064459 | Deep neural networks achieve remarkable performance in many computer vision tasks. Most state-of-the-art (SOTA) semantic segmentation and object detection approaches reuse neural network architectures designed for image classification as the backbone, commonly pre-trained on ImageNet. However, performance gains can be achieved by designing network architectures specifically for detection and segmentation, as shown by recent neural architecture search (NAS) research for detection and segmentation. One major challenge though, is that ImageNet pre-training of the search space representation (a.k.a. super network) or the searched networks incurs huge computational cost. In this paper, we propose a Fast Neural Network Adaptation (FNA) method, which can adapt both the architecture and parameters of a seed network (e.g. a high performing manually designed backbone) to become a network with different depth, width, or kernels via a Parameter Remapping technique, making it possible to utilize NAS for detection/segmentation tasks a lot more efficiently. In our experiments, we conduct FNA on MobileNetV2 to obtain new networks for both segmentation and detection that clearly out-perform existing networks designed both manually and by NAS. The total computation cost of FNA is significantly less than SOTA segmentation/detection NAS approaches: 1737× less than DPC, 6.8× less than Auto-DeepLab and 7.4× less than DetNAS. The code is available at https://github.com/JaminFong/FNA. * Equal contributions. † The work is performed during an internship at Horizon Robotics. | Published as a conference paper at ICLR 2020 FAST NEURAL NETWORK ADAPTATION VIA PARAME- TER REMAPPING AND ARCHITECTURE SEARCH |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.