_id stringlengths 4 10 | text stringlengths 0 18.4k | title stringlengths 0 8.56k |
|---|---|---|
d232222444 | We accelerate deep reinforcement learning-based training in visually complex 3D environments by two orders of magnitude over prior work, realizing end-to-end training speeds of over 19,000 frames of experience per second on a single GPU and up to 72,000 frames per second on a single eight-GPU machine. The key idea of our approach is to design a 3D renderer and embodied navigation simulator around the principle of "batch simulation": accepting and executing large batches of requests simultaneously. Beyond exposing large amounts of work at once, batch simulation allows implementations to amortize in-memory storage of scene assets, rendering work, data loading, and synchronization costs across many simulation requests, dramatically improving the number of simulated agents per GPU and overall simulation throughput. To balance DNN inference and training costs with faster simulation, we also build a computationally efficient policy DNN that maintains high task performance, and modify training algorithms to maintain sample efficiency when training with large mini-batches. By combining batch simulation and DNN performance optimizations, we demonstrate that PointGoal navigation agents can be trained in complex 3D environments on a single GPU in 1.5 days to 97% of the accuracy of agents trained on a prior state-of-the-art system using a 64-GPU cluster over three days. We provide open-source reference implementations of our batch 3D renderer and simulator to facilitate incorporation of these ideas into RL systems. | Published as a conference paper at ICLR 2021 LARGE BATCH SIMULATION FOR DEEP REINFORCEMENT LEARNING |
d212415013 | Many neural network pruning algorithms proceed in three steps: train the network to completion, remove unwanted structure to compress the network, and retrain the remaining structure to recover lost accuracy. The standard retraining technique, fine-tuning, trains the unpruned weights from their final trained values using a small fixed learning rate. In this paper, we compare fine-tuning to alternative retraining techniques. Weight rewinding (as proposed by Frankle et al.(2019)), rewinds unpruned weights to their values from earlier in training and retrains them from there using the original training schedule. Learning rate rewinding (which we propose) trains the unpruned weights from their final values using the same learning rate schedule as weight rewinding. Both rewinding techniques outperform fine-tuning, forming the basis of a network-agnostic pruning algorithm that matches the accuracy and compression ratios of several more network-specific state-of-the-art techniques. | Published as a conference paper at ICLR 2020 COMPARING REWINDING AND FINE-TUNING IN NEURAL NETWORK PRUNING |
d15336613 | Convolutional neural networks[3]have proven useful in many domains, including computer vision[1,4,5], audio processing[6,7]and natural language processing[8]. These powerful models come at great cost in training time, however. Currently, long training periods make experimentation difficult and time consuming.In this work, we consider a standard architecture [1] trained on the Imagenet dataset [2] for classification and investigate methods to speed convergence by parallelizing training across multiple GPUs. In this work, we used up to 4 NVIDIA TITAN GPUs with 6GB of RAM. While our experiments are performed on a single server, our GPUs have disjoint memory spaces, and just as in the distributed setting, communication overheads are an important consideration. Unlike previous work [9, 10, 11], we do not aim to improve the underlying optimization algorithm. Instead, we isolate the impact of parallelism, while using standard supervised back-propagation and synchronous mini-batch stochastic gradient descent.We consider two basic approaches: data and model parallelism[9]. In data parallelism, the minibatch is split across several GPUs as shown infig. 3. Each GPU is responsible for computing gradients with respect to all model parameters, but does so using a subset of the samples in the mini-batch. This is the most straightforward parallelization method, but it requires considerable communication between GPUs, since each GPU must communicate both gradients and parameter values on every update step. Also, each GPU must use a large number of samples to effectively utilize the highly parallel device; thus, the mini-batch size effectively gets multiplied by the number of GPUs, hampering convergence. In our implementation, we find a speed-up of 1.5 times moving from 1 to 2 GPUs in the data parallel framework (when using 2 GPUs, each gets assigned a minibatch of size 128). This experiment used the same architecture, network set up and dataset described in Krizhevsky et al.[1].Model parallelism, on the other hand, consists of splitting an individual network's computation across multiple GPUs[9,12]. An example is shown infig. 4. For instance, a convolutional layer with N filters can be run on two GPUs, each of which convolves its input with N/2 filters. In their seminal work, Krizhevsky et al.[1]further customized the architecture of the network to better leverage model parallelism: the architecture consists of two "columns" each allocated on one GPU. | Multi-GPU Training of ConvNets |
d5161803 | 6[cs.LG]
We define Recurrent Gaussian Processes (RGP) models, a general family of Bayesian nonparametric models with recurrent GP priors which are able to learn dynamical patterns from sequential data.Similar to Recurrent Neural Networks (RNNs), RGPs can have different formulations for their internal states, distinct inference methods and be extended with deep structures.In such context, we propose a novel deep RGP model whose autoregressive states are latent, thereby performing representation and dynamical learning simultaneously.To fully exploit the Bayesian nature of the RGP model we develop the Recurrent Variational Bayes (REVARB) framework, which enables efficient inference and strong regularization through coherent propagation of uncertainty across the RGP layers and states.We also introduce a RGP extension where variational parameters are greatly reduced by being reparametrized through RNN-based sequential recognition models.We apply our model to the tasks of nonlinear system identification and human motion modeling.The promising obtained results indicate that our RGP model maintains its highly flexibility while being able to avoid overfitting and being applicable even when larger datasets are not availab | RECURRENT GAUSSIAN PROCESSES
24 Feb 2016
César Lincoln cesarlincoln@terr |
d244709713 | Simulation-based inference with conditional neural density estimators is a powerful approach to solving inverse problems in science. However, these methods typically treat the underlying forward model as a black box, with no way to exploit geometric properties such as equivariances. Equivariances are common in scientific models, however integrating them directly into expressive inference networks (such as normalizing flows) is not straightforward. We here describe an alternative method to incorporate equivariances under joint transformations of parameters and data. Our method-called group equivariant neural posterior estimation (GNPE)-is based on self-consistently standardizing the "pose" of the data while estimating the posterior over parameters. It is architecture-independent, and applies both to exact and approximate equivariances. As a real-world application, we use GNPE for amortized inference of astrophysical binary black hole systems from gravitationalwave observations. We show that GNPE achieves state-of-the-art accuracy while reducing inference times by three orders of magnitude.arXiv:2111.13139v2 [cs.LG] 30 May 2023Published as a conference paper at ICLR 2022 Hanford Livingston s t a n d a r d i z e d d i r e c t i o n VirgoFigure 1: By standardizing the source sky position, a GW signal can be made to arrive at the same time in all three LIGO/Virgo detectors. However, since this also changes the projection of the signal onto the detectors, it defines only an approximate equivariance. Nevertheless, our proposed GNPE algorithm simplifies inference by simultaneously inferring and standardizing the incident direction.Training an inference network for any x ∼ p(x) can nevertheless present challenges due to the large number of training samples and powerful networks required. The present study is motivated by the problem of gravitational-wave (GW) data analysis. Here the task is to infer properties of astrophysical black-hole mergers based on GW signals observed at the LIGO and Virgo observatories on Earth.Due to the complexity of signal models, it has previously not been possible to train networks to estimate posteriors to the same accuracy as conventional likelihood-based methods(Veitch et al., 2015;Ashton et al., 2019). The GW posterior, however, is equivariant 1 under an overall change in the time of arrival of the data. It is also approximately equivariant under a joint change in the sky position and (by triangulation) individual shifts in the arrival times in each detector(Fig. 1). If we could constrain these parameters a priori, we could therefore apply time shifts to align the detector data and simplify the inference task for the remaining parameters.More generally, we consider forward models with known equivariances under group transformations applied jointly to data and parameters. Our aim is to exploit this knowledge to standardize the pose of the data 2 and simplify analysis. The obvious roadblock is that the pose is contained in the set of parameters θ and is therefore unknown prior to inference. Here we describe group equivariant neural posterior estimation (GNPE), a method to self-consistently infer parameters and standardize the pose. The basic approach is to introduce a proxy for the pose-a blurred version-on which one conditions the posterior. The pose of the data is then transformed based on the proxy, placing it in a band about the standard value, and resulting in an easier inference task. Finally, the joint posterior over θ and the pose proxy can be sampled at inference time using Gibbs sampling.The standard method to incorporating equivariances is to integrate them directly into network architectures, e.g., to use convolutional networks for translational equivariances. Although these approaches can be highly effective, they impose design constraints on network architectures. For GWs, for example, we use specialized embedding networks to extract signal waveforms from frequencydomain data, as well as expressive normalizing flows to estimate the posterior-neither of which is straightforward to make explicitly equivariant. We also have complex equivariance connections between subsets of parameters and data, including approximate equivariances. The GNPE algorithm is extremely general: it is architecture-independent, it applies whether equivariances are exact or approximate, and it allows for arbitrary equivariance relations between parameters and data.We discuss related work in Sec. 2 and describe the GNPE algorithm in Sec. 3. In Sec. 4 we apply GNPE to a toy example with exact translational equivariance, showing comparable simulation efficiency to NPE combined with a convolutional network. In Sec. 5 we show that standard NPE does not achieve adequate accuracy for GW parameter inference, even with an essentially unlimited number of simulations. In contrast, GNPE achieves highly accurate posteriors at a computational cost three orders of magnitude lower than bespoke MCMC approaches(Veitch et al., 2015). The present paper describes the GNPE method which we developed for GW analysis (Dax et al., 2021), and extends it to general equivariance transformations which makes it applicable to a wide range of problems. A detailed description of GW results is presented in Dax et al. (2021). | Published as a conference paper at ICLR 2022 GROUP EQUIVARIANT NEURAL POSTERIOR ESTIMATION |
d252734772 | Though end-to-end neural approaches have recently been dominating NLP tasks in both performance and ease-of-use, they lack interpretability and robustness. We propose BINDER, a training-free neural-symbolic framework that maps the task input to a program, which (1) allows binding a unified API of language model (LM) functionalities to a programming language (e.g., SQL, Python) to extend its grammar coverage and thus tackle more diverse questions, (2) adopts an LM as both the program parser and the underlying model called by the API during execution, and(3)requires only a few in-context exemplar annotations. Specifically, we employ GPT-3 Codex as the LM. In the parsing stage, with only a few incontext exemplars, Codex is able to identify the part of the task input that cannot be answerable by the original programming language, correctly generate API calls to prompt Codex to solve the unanswerable part, and identify where to place the API calls while being compatible with the original grammar. In the execution stage, Codex can perform versatile functionalities (e.g., commonsense QA, information extraction) given proper prompts in the API calls. BINDER achieves state-of-the-art results on WIKITABLEQUESTIONS and TABFACT datasets, with explicit output programs that benefit human debugging. Note that previous best systems are all finetuned on tens of thousands of task-specific samples, while BINDER only uses dozens of annotations as in-context exemplars without any training. Our code is available at https://github.com/hkunlp/binder 1 . * Equal contribution. Authors in alphabetical order. Work mainly done at the University of Hong Kong. 1 More resources at https://lm-code-binder.github.io/. | Published as a conference paper at ICLR 2023 BINDING LANGUAGE MODELS IN SYMBOLIC LANGUAGES |
d14047116 | Generative Stochastic Networks (GSNs) have been recently introduced as an alternative to traditional probabilistic modeling: instead of parametrizing the data distribution directly, one parametrizes a transition operator for a Markov chain whose stationary distribution is an estimator of the data generating distribution. The result of training is therefore a machine that generates samples through this Markov chain. However, the previously introduced GSN consistency theorems suggest that in order to capture a wide class of distributions, the transition operator in general should be multimodal, something that has not been done before this paper. We introduce for the first time multimodal transition distributions for GSNs, in particular using models in the NADE family (Neural Autoregressive Density Estimator) as output distributions of the transition operator. A NADE model is related to an RBM (and can thus model multimodal distributions) but its likelihood (and likelihood gradient) can be computed easily. The parameters of the NADE are obtained as a learned function of the previous state of the learned Markov chain. Experiments clearly illustrate the advantage of such multimodal transition distributions over unimodal GSNs. | Multimodal Transitions for Generative Stochastic Networks |
d248239666 | The problem of processing very long time-series data (e.g., a length of more than 10,000) is a long-standing research problem in machine learning. Recently, one breakthrough, called neural rough differential equations (NRDEs), has been proposed and has shown that it is able to process such data. Their main concept is to use the log-signature transform, which is known to be more efficient than the Fourier transform for irregular long time-series, to convert a very long timeseries sample into a relatively shorter series of feature vectors. However, the logsignature transform causes non-trivial spatial overheads. To this end, we present the method of LOweR-Dimensional embedding of log-signature (LORD), where we define an NRDE-based autoencoder to implant the higher-depth log-signature knowledge into the lower-depth log-signature. We show that the encoder successfully combines the higher-depth and the lower-depth log-signature knowledge, which greatly stabilizes the training process and increases the model accuracy. In our experiments with benchmark datasets, the improvement ratio by our method is up to 75% in terms of various classification and forecasting evaluation metrics.arXiv:2204.08781v1 [cs.LG] 19 Apr 2022Published as a conference paper at ICLR 2022 t 0 t 1 t 2 t N ... | LORD: LOWER-DIMENSIONAL EMBEDDING OF LOG- SIGNATURE IN NEURAL ROUGH DIFFERENTIAL EQUA- TIONS |
d3602416 | We relate the minimax game of generative adversarial networks (GANs) to finding the saddle points of the Lagrangian function for a convex optimization problem, where the discriminator outputs and the distribution of generator outputs play the roles of primal variables and dual variables, respectively. This formulation shows the connection between the standard GAN training process and the primal-dual subgradient methods for convex optimization. The inherent connection does not only provide a theoretical convergence proof for training GANs in the function space, but also inspires a novel objective function for training. The modified objective function forces the distribution of generator outputs to be updated along the direction according to the primal-dual subgradient methods. A toy example shows that the proposed method is able to resolve mode collapse, which in this case cannot be avoided by the standard GAN or Wasserstein GAN. Experiments on both Gaussian mixture synthetic data and real-world image datasets demonstrate the performance of the proposed method on generating diverse samples. * The first two authors have equal contributions. | Published as a conference paper at ICLR 2018 TRAINING GENERATIVE ADVERSARIAL NETWORKS VIA PRIMAL-DUAL SUBGRADIENT METHODS: A LAGRANGIAN PERSPECTIVE |
d257232537 | Multimodal few-shot learning is challenging due to the large domain gap between vision and language modalities. Existing methods are trying to communicate visual concepts as prompts to frozen language models, but rely on hand-engineered task induction to reduce the hypothesis space. To make the whole process learnable, we introduce a multimodal meta-learning approach. Specifically, our approach decomposes the training of the model into a set of related multimodal few-shot tasks. We define a meta-mapper network, acting as a meta-learner, to efficiently bridge frozen large-scale vision and language models and leverage their already learned capacity. By updating the learnable parameters only of the meta-mapper, it learns to accrue shared meta-knowledge among these tasks. Thus, it can rapidly adapt to newly presented samples with only a few gradient updates. Importantly, it induces the task in a completely data-driven manner, with no need for a hand-engineered task induction. We evaluate our approach on recently proposed multimodal few-shot benchmarks, measuring how rapidly the model can bind novel visual concepts to words and answer visual questions by observing only a limited set of labeled examples. The experimental results show that our meta-learning approach outperforms the baseline across multiple datasets and various training settings while being computationally more efficient. | Published as a conference paper at ICLR 2023 META LEARNING TO BRIDGE VISION AND LANGUAGE MODELS FOR MULTIMODAL FEW-SHOT LEARNING |
d209334533 | Predicting outcomes and planning interactions with the physical world are longstanding goals for machine learning. A variety of such tasks involves continuous physical systems, which can be described by partial differential equations (PDEs) with many degrees of freedom. Existing methods that aim to control the dynamics of such systems are typically limited to relatively short time frames or a small number of interaction parameters. We present a novel hierarchical predictorcorrector scheme which enables neural networks to learn to understand and control complex nonlinear physical systems over long time frames. We propose to split the problem into two distinct tasks: planning and control. To this end, we introduce a predictor network that plans optimal trajectories and a control network that infers the corresponding control parameters. Both stages are trained end-to-end using a differentiable PDE solver. We demonstrate that our method successfully develops an understanding of complex physical systems and learns to control them for tasks involving PDEs such as the incompressible Navier-Stokes equations. A multiple shooting algorithm for direct solution of optimal control problems. | Published as a conference paper at ICLR 2020 LEARNING TO CONTROL PDES WITH DIFFERENTIABLE PHYSICS |
d11383178 | The variational autoencoder (VAE; ) is a recently proposed generative model pairing a top-down generative network with a bottom-up recognition network which approximates posterior inference. It typically makes strong assumptions about posterior inference, for instance that the posterior distribution is approximately factorial, and that its parameters can be approximated with nonlinear regression from the observations. As we show empirically, the VAE objective can lead to overly simplified representations which fail to use the network's entire modeling capacity. We present the importance weighted autoencoder (IWAE), a generative model with the same architecture as the VAE, but which uses a strictly tighter log-likelihood lower bound derived from importance weighting. In the IWAE, the recognition network uses multiple samples to approximate the posterior, giving it increased flexibility to model complex posteriors which do not fit the VAE modeling assumptions. We show empirically that IWAEs learn richer latent space representations than VAEs, leading to improved test log-likelihood on density estimation benchmarks. | IMPORTANCE WEIGHTED AUTOENCODERS |
d249209875 | Offline reinforcement learning (RL) aims at learning an optimal strategy using a pre-collected dataset without further interactions with the environment. While various algorithms have been proposed for offline RL in the previous literature, the minimax optimality has only been (nearly) established for tabular Markov decision processes (MDPs). In this paper, we focus on offline RL with linear function approximation and propose a new pessimism-based algorithm for offline linear MDP. At the core of our algorithm is the uncertainty decomposition via a reference function, which is new in the literature of offline RL under linear function approximation. Theoretical analysis demonstrates that our algorithm can match the performance lower bound up to logarithmic factors. We also extend our techniques to the two-player zero-sum Markov games (MGs), and establish a new performance lower bound for MGs, which tightens the existing result, and verifies the nearly minimax optimality of the proposed algorithm. To the best of our knowledge, these are the first computationally efficient and nearly minimax optimal algorithms for offline single-agent MDPs and MGs with linear function approximation. * The first two authors contribute equally. | NEARLY MINIMAX OPTIMAL OFFLINE REINFORCE- MENT LEARNING WITH LINEAR FUNCTION APPROXI- MATION: SINGLE-AGENT MDP AND MARKOV GAME |
d258212652 | Decoupling representation learning and classifier learning has been shown to be effective in classification with long-tailed data. There are two main ingredients in constructing a decoupled learning scheme; 1) how to train the feature extractor for representation learning so that it provides generalizable representations and 2) how to re-train the classifier that constructs proper decision boundaries by handling class imbalances in long-tailed data. In this work, we first apply Stochastic Weight Averaging (SWA), an optimization technique for improving the generalization of deep neural networks, to obtain better generalizing feature extractors for long-tailed classification. We then propose a novel classifier re-training algorithm based on stochastic representation obtained from the SWA-Gaussian, a Gaussian perturbed SWA, and a self-distillation strategy that can harness the diverse stochastic representations based on uncertainty estimates to build more robust classifiers. Extensive experiments on CIFAR10/100-LT, ImageNet-LT, and iNaturalist-2018 benchmarks show that our proposed method improves upon previous methods both in terms of prediction accuracy and uncertainty estimation. | Published as a conference paper at ICLR 2023 DECOUPLED TRAINING FOR LONG-TAILED CLASSIFI- CATION WITH STOCHASTIC REPRESENTATIONS |
d247158827 | It is well-known that deep neural networks (DNNs) are susceptible to adversarial attacks, exposing a severe fragility of deep learning systems. As the result, adversarial training (AT) method, by incorporating adversarial examples during training, represents a natural and effective approach to strengthen the robustness of a DNN-based classifier. However, most AT-based methods, notably PGD-AT and TRADES, typically seek a pointwise adversary that generates the worst-case adversarial example by independently perturbing each data sample, as a way to "probe" the vulnerability of the classifier. Arguably, there are unexplored benefits in considering such adversarial effects from an entire distribution. To this end, this paper presents a unified framework that connects Wasserstein distributional robustness with current state-of-the-art AT methods. We introduce a new Wasserstein cost function and a new series of risk functions, with which we show that standard AT methods are special cases of their counterparts in our framework. This connection leads to an intuitive relaxation and generalization of existing AT methods and facilitates the development of a new family of distributional robustness AT-based algorithms. Extensive experiments show that our distributional robustness AT algorithms robustify further their standard AT counterparts in various settings. 1 1 Our code is available at https://github.com/tuananhbui89/Unified-Distributional-Robustness | A UNIFIED WASSERSTEIN DISTRIBUTIONAL ROBUST- NESS FRAMEWORK FOR ADVERSARIAL TRAINING |
d256615635 | Recent vision transformer based video models mostly follow the "image pretraining then finetuning" paradigm and have achieved great success on multiple video benchmarks. However, full finetuning such a video model could be computationally expensive and unnecessary, given the pre-trained image transformer models have demonstrated exceptional transferability. In this work, we propose a novel method to Adapt pre-trained Image Models (AIM) for efficient video understanding. By freezing the pre-trained image model and adding a few lightweight Adapters, we introduce spatial adaptation, temporal adaptation and joint adaptation to gradually equip an image model with spatiotemporal reasoning capability. We show that our proposed AIM can achieve competitive or even better performance than prior arts with substantially fewer tunable parameters on four video action recognition benchmarks. Thanks to its simplicity, our method is also generally applicable to different image pre-trained models, which has the potential to leverage more powerful image foundation models in the future. The project webpage is https://adapt-image-models.github.io/. * Work done during an internship at Amazon Web Services. | Published as a conference paper at ICLR 2023 AIM: ADAPTING IMAGE MODELS FOR EFFICIENT VIDEO ACTION RECOGNITION |
d214309387 | We introduce a deep recurrent neural network architecture that approximates visual cortical circuits(Mély et al., 2018). We show that this architecture, which we refer to as the γ-Net, learns to solve contour detection tasks with better sample efficiency than state-of-the-art feedforward networks, while also exhibiting a classic perceptual illusion, known as the orientation-tilt illusion. Correcting this illusion significantly reduces γ-Net contour detection accuracy by driving it to prefer lowlevel edges over high-level object boundary contours. Overall, our study suggests that the orientation-tilt illusion is a byproduct of neural circuits that help biological visual systems achieve robust and efficient contour detection, and that incorporating these circuits in artificial neural networks can improve computer vision. † These authors contributed equally to this work.Published as a conference paper at ICLR 2020 possibility and demonstrate that the orientation-tilt illusion reflects neural strategies optimized for object contour detection.Contributions We introduce the γ-Net, a trainable and hierarchical extension of the neural circuit of Mély et al.(2018), which explains contextual illusions. (i) The γ-Net is more sample efficient than state-of-the-art convolutional architectures on two separate contour detection tasks. (ii) Similar to humans but not state-of-the-art contour detection models, the γ-Net exhibits an orientation-tilt illusion after being optimized for contour detection. This illusion emerges from its preference for high-level object-boundary contours over low-level edges, indicating that neural circuits involved in contextual illusions also support sample-efficient solutions to contour detection tasks.RELATED WORKModeling the visual system Convolutional neural networks (CNNs) are often considered the de facto "standard model" of vision. CNNs and their extensions represent the state of the art for most computer vision applications with performance approaching -and sometimes exceeding -human observers on certain visual recognition tasks (He et al. | Published as a conference paper at ICLR 2020 RECURRENT NEURAL CIRCUITS FOR CONTOUR DETECTION |
d252568221 | In the literature on game-theoretic equilibrium finding, focus has mainly been on solving a single game in isolation. In practice, however, strategic interactions-ranging from routing problems to online advertising auctions-evolve dynamically, thereby leading to many similar games to be solved. To address this gap, we introduce meta-learning for equilibrium finding and learning to play games. We establish the first meta-learning guarantees for a variety of fundamental and well-studied classes of games, including two-player zero-sum games, general-sum games, and Stackelberg games. In particular, we obtain rates of convergence to different game-theoretic equilibria that depend on natural notions of similarity between the sequence of games encountered, while at the same time recovering the known single-game guarantees when the sequence of games is arbitrary. Along the way, we prove a number of new results in the single-game regime through a simple and unified framework, which may be of independent interest. Finally, we evaluate our meta-learning algorithms on endgames faced by the poker agent Libratus against top human professionals. The experiments show that games with varying stack sizes can be solved significantly faster using our meta-learning techniques than by solving them separately, often by an order of magnitude. | Meta-Learning in Games * |
d238407888 | Neural collapse is a highly symmetric geometry of neural networks that emerges during the terminal phase of training, with profound implications on the generalization performance and robustness of the trained networks. To understand how the last-layer features and classifiers exhibit this recently discovered implicit bias, in this paper, we introduce a surrogate model called the unconstrained layer-peeled model (ULPM). We prove that gradient flow on this model converges to critical points of a minimum-norm separation problem exhibiting neural collapse in its global minimizer. Moreover, we show that the ULPM with the cross-entropy loss has a benign global landscape for its loss function, which allows us to prove that all the critical points are strict saddle points except the global minimizers that exhibit the neural collapse phenomenon. Empirically, we show that our results also hold during the training of neural networks in real-world tasks when explicit regularization or weight decay is not used. , et al. Language models are few-shot learners. arXiv preprint arXiv:2005.14165, 2020.Lenaic Chizat and Francis Bach. Implicit bias of gradient descent for wide two-layer neural networks trained with the logistic loss. . Escaping from saddle points-online stochastic gradient for tensor decomposition. In Conference on learning theory, pp. 797-842. PMLR, 2015.Rong Ge, Jason D Lee, and Tengyu Ma. Matrix completion has no spurious local minimum. arXiv preprint arXiv:1605.07272, 2016. | Published as a conference paper at ICLR 2022 AN UNCONSTRAINED LAYER-PEELED PERSPECTIVE ON NEURAL COLLAPSE |
d211011309 | Network quantization is one of the most hardware friendly techniques to enable the deployment of convolutional neural networks (CNNs) on low-power mobile devices. Recent network quantization techniques quantize each weight kernel in a convolutional layer independently for higher inference accuracy, since the weight kernels in a layer exhibit different variances and hence have different amounts of redundancy. The quantization bitwidth or bit number (QBN) directly decides the inference accuracy, latency, energy and hardware overhead. To effectively reduce the redundancy and accelerate CNN inferences, various weight kernels should be quantized with different QBNs. However, prior works use only one QBN to quantize each convolutional layer or the entire CNN, because the design space of searching a QBN for each weight kernel is too large. The hand-crafted heuristic of the kernel-wise QBN search is so sophisticated that domain experts can obtain only sub-optimal results. It is difficult for even deep reinforcement learning (DRL) Deep Deterministic Policy Gradient (DDPG)-based agents to find a kernel-wise QBN configuration that can achieve reasonable inference accuracy. In this paper, we propose a hierarchical-DRL-based kernel-wise network quantization technique, AutoQ, to automatically search a QBN for each weight kernel, and choose another QBN for each activation layer. Compared to the models quantized by the state-of-the-art DRL-based schemes, the same models quantized by AutoQ reduce the inference latency by 54.06%, and decrease the inference energy consumption by 50.69% averagely, while achieving the same inference accuracy. | Published as a conference paper at ICLR 2020 AUTOQ: AUTOMATED KERNEL-WISE NEURAL NETWORK QUANTIZATION * |
d257756989 | In-Context Learning (ICL), which formulates target tasks as prompt completion conditioned on in-context demonstrations, has become the prevailing utilization of LLMs. In this paper, we first disclose an actual predicament for this typical usage that it can not scale up with training data due to context length restriction. Besides, existing works have shown that ICL also suffers from various biases and requires delicate calibration treatment. To address both challenges, we advocate a simple and effective solution, kNN Prompting, which first queries LLM with training data for distributed representations, then predicts test instances by simply referring to nearest neighbors. We conduct comprehensive experiments to demonstrate its two-fold superiority: 1) Calibration-Free: kNN Prompting does not directly align LLM output distribution with task-specific label space, instead leverages such distribution to align test and training instances. It significantly outperforms state-of-the-art calibration-based methods under comparable few-shot scenario. 2) Beyond-Context: kNN Prompting can further scale up effectively with as many training data as are available, continually bringing substantial improvements. The scaling trend holds across 10 orders of magnitude ranging from 2 shots to 1024 shots as well as different LLMs scales ranging from 0.8B to 30B. It successfully bridges data scaling into model scaling, and brings new potentials for the gradient-free paradigm of LLM deployment. Code is publicly available 1 .Published as a conference paper at ICLR 2023 prompt is constructed by concatenation of several training examples and a test instance, and the prediction is obtained by mapping the LLM word continuations back to label space.It is widely investigated and acknowledged that modern neural networks generally perform better w.r.t. increased training data. Specifically, there exists a power law between expected model performance and available data scale (Hestness et al., 2017;Rosenfeld et al., 2020). For ICL, it is also empirically observed that the performance continually improves when more training examples are prepended into the prompt(Brown et al., 2020). However, such improvements are quickly prevented by the predicament of context length restriction, as language models are designed and trained to only process sequences within a fixed length, which is in fact 1024 or 2048 tokens. In order to utilize more training data, several works try to select the most relevant examples to compose the prompt before querying LLM (Liu et al., 2022b;Rubin et al., 2022), but still only in-context examples can actually participate the LLM inference while most training data are discarded beforehand, thus providing marginal data scaling benefits. Besides, their reliance on external retriever also incurs further complications. As a consequence, such a situation poses a serious challenge for many practical scenarios where more than a few training data are available. | Published as a conference paper at ICLR 2023 kNN PROMPTING: BEYOND-CONTEXT LEARNING WITH CALIBRATION-FREE NEAREST NEIGHBOR IN- FERENCE |
d252780278 | Demonstrations and natural language instructions are two common ways to specify and teach robots novel tasks. However, for many complex tasks, a demonstration or language instruction alone contains ambiguities, preventing tasks from being specified clearly. In such cases, a combination of both a demonstration and an instruction more concisely and effectively conveys the task to the robot than either modality alone. To instantiate this problem setting, we train a single multi-task policy on a few hundred challenging robotic pick-and-place tasks and propose DeL-TaCo (Joint Demo-Language Task Conditioning), a method for conditioning a robotic policy on task embeddings comprised of two components: a visual demonstration and a language instruction. By allowing these two modalities to mutually disambiguate and clarify each other during novel task specification, DeL-TaCo (1) substantially decreases the teacher effort needed to specify a new task and (2) achieves better generalization performance on novel objects and instructions over previous task-conditioning methods. To our knowledge, this is the first work to show that simultaneously conditioning a multi-task robotic manipulation policy on both demonstration and language embeddings improves sample efficiency and generalization over conditioning on either modality alone. See additional materials at https://deltaco-robot.github.io. | Published as a conference paper at ICLR 2023 USING BOTH DEMONSTRATIONS AND LANGUAGE IN- STRUCTIONS TO EFFICIENTLY LEARN ROBOTIC TASKS |
d2222933 | One major challenge in training Deep Neural Networks is preventing overfitting. Many techniques such as data augmentation and novel regularizers such as Dropout have been proposed to prevent overfitting without requiring a massive amount of training data. In this work, we propose a new regularizer called DeCov which leads to significantly reduced overfitting (as indicated by the difference between train and val performance), and better generalization. Our regularizer encourages diverse or non-redundant representations in Deep Neural Networks by minimizing the cross-covariance of hidden activations. This simple intuition has been explored in a number of past works but surprisingly has never been applied as a regularizer in supervised learning. Experiments across a range of datasets and network architectures show that this loss always reduces overfitting while almost always maintaining or increasing generalization performance and often improving performance over Dropout. | Published as a conference paper at ICLR 2016 REDUCING OVERFITTING IN DEEP NETWORKS BY DECORRELATING REPRESENTATIONS |
d256598058 | Most graph neural networks follow the message passing mechanism. However, it faces the over-smoothing problem when multiple times of message passing is applied to a graph, causing indistinguishable node representations and prevents the model to effectively learn dependencies between farther-away nodes. On the other hand, features of neighboring nodes with different labels are likely to be falsely mixed, resulting in the heterophily problem. In this work, we propose to order the messages passing into the node representation, with specific blocks of neurons targeted for message passing within specific hops. This is achieved by aligning the hierarchy of the rooted-tree of a central node with the ordered neurons in its node representation. Experimental results on an extensive set of datasets show that our model can simultaneously achieve the state-of-the-art in both homophily and heterophily settings, without any targeted design. Moreover, its performance maintains pretty well while the model becomes really deep, effectively preventing the over-smoothing problem. Finally, visualizing the gating vectors shows that our model learns to behave differently between homophily and heterophily settings, providing an explainable graph neural model. | Published as a conference paper at ICLR 2023 ORDERED GNN: ORDERING MESSAGE PASSING TO DEAL WITH HETEROPHILY AND OVER-SMOOTHING |
d239768902 | Despite being able to capture a range of features of the data, high accuracy models trained with supervision tend to make similar predictions. This seemingly implies that high-performing models share similar biases regardless of training methodology, which would limit ensembling benefits and render low-accuracy models as having little practical use. Against this backdrop, recent work has developed quite different training techniques, such as large-scale contrastive learning, yielding competitively high accuracy on generalization and robustness benchmarks. This motivates us to revisit the assumption that models necessarily learn similar functions. We conduct a large-scale empirical study of models across hyper-parameters, architectures, frameworks, and datasets. We find that model pairs that diverge more in training methodology display categorically different generalization behavior, producing increasingly uncorrelated errors. We show these models specialize in subdomains of the data, leading to higher ensemble performance: with just 2 models (each with ImageNet accuracy76.5%), we can create ensembles with 83.4% (+7% boost). Surprisingly, we find that even significantly low-accuracy models can be used to improve high-accuracy models. Finally, we show diverging training methodology yield representations that capture overlapping (but not supersetting) feature sets which, when combined, lead to increased downstream performance.arXiv:2110.12899v3 [cs.LG] 25 Apr 2022Published as a conference paper at ICLR 2022To settle these questions, we conduct a systematic empirical study of 82 models, which we train or collect, across hyper-parameters, architectures, objective functions, and datasets, including the latest high performing models CLIP, ALIGN, SimCLR, BiT, ViT-G/14, and MPL. In addition to using different techniques, these new models were trained on data collected very differently, allowing us to probe the effect of both training objective, as well as pre-training data. We categorize these models based on how their training methodologies diverge from a typical, base model and show:1. Model pairs that diverge more in training methodology (in order: reinitializations ) hyperparameters ) architectures ) frameworks ) datasets) produce increasingly uncorrelated errors. 2. Ensemble performance increases as error correlation decreases, due to higher ensemble efficiency. The most typical ImageNet model (ResNet-50, 76.5%), and its most different counterpart (ALIGN-ZS, 75.5%) yield 83.4% accuracy when ensembled, a +7% boost. 3. Contrastively-learned models display categorically different generalization behavior, specializing in subdomains of the data, which explains the higher ensembling efficiency. We show CLIP-S specializes in antropogenic images, whereas ResNet-50 excels in nature images. 4. Surprisingly, we find that low-accuracy models can be useful if they are trained differently enough. By combining a high-accuracy model (BiT-1k, 82.9%) with only low-accuracy models (max individual acc. 77.4%), we can create ensembles that yield as much as 86.7%. 5. Diverging training methodology yield representations that capture overlapping (but not supersetting) feature sets which, when concatenated, lead to increased downstream performance (91.4% on Pascal VOC, using models with max individual accuracy 90.7%). | Published as a conference paper at ICLR 2022 NO ONE REPRESENTATION TO RULE THEM ALL: OVERLAPPING FEATURES OF TRAINING METHODS |
d209461008 | The allocation of computation resources in the backbone is a crucial issue in object detection. However, classification allocation pattern is usually adopted directly to object detector, which is proved to be sub-optimal. In order to reallocate the engaged computation resources in a more efficient way, we present CR-NAS (Computation Reallocation Neural Architecture Search) that can learn computation reallocation strategies across different feature resolution and spatial position diectly on the target detection dataset. A two-level reallocation space is proposed for both stage and spatial reallocation. A novel hierarchical search procedure is adopted to cope with the complex search space. We apply CR-NAS to multiple backbones and achieve consistent improvements. Our CR-ResNet50 and CR-MobileNetV2 outperforms the baseline by 1.9% and 1.7% COCO AP respectively without any additional computation budget. The models discovered by CR-NAS can be equiped to other powerful detection neck/head and be easily transferred to other dataset, e.g. PASCAL VOC, and other vision tasks, e.g. instance segmentation. Our CR-NAS can be used as a plugin to improve the performance of various networks, which is demanding. | Published as a conference paper at ICLR 2020 COMPUTATION REALLOCATION FOR OBJECT DETECTION |
d256827203 | Mixed-integer linear programming (MILP) is widely employed for modeling combinatorial optimization problems. In practice, similar MILP instances with only coefficient variations are routinely solved, and machine learning (ML) algorithms are capable of capturing common patterns across these MILP instances. In this work, we combine ML with optimization and propose a novel predict-and-search framework for efficiently identifying high-quality feasible solutions. Specifically, we first utilize graph neural networks to predict the marginal probability of each variable, and then search for the best feasible solution within a properly defined ball around the predicted solution. We conduct extensive experiments on public datasets, and computational results demonstrate that our proposed framework achieves 51.1% and 9.9% performance improvements to MILP solvers SCIP and Gurobi on primal gaps, respectively. | A GNN-GUIDED PREDICT-AND-SEARCH FRAME- WORK FOR MIXED-INTEGER LINEAR PROGRAMMING |
d245906452 | The architecture and the parameters of neural networks are often optimized independently, which requires costly retraining of the parameters whenever the architecture is modified. In this work we instead focus on growing the architecture without requiring costly retraining. We present a method that adds new neurons during training without impacting what is already learned, while improving the training dynamics. We achieve the latter by maximizing the gradients of the new weights and efficiently find the optimal initialization by means of the singular value decomposition (SVD). We call this technique Gradient Maximizing Growth (GradMax) and demonstrate its effectiveness in variety of vision tasks and architectures 1 . | Published as a conference paper at ICLR 2022 GRADMAX: GROWING NEURAL NETWORKS USING GRADIENT INFORMATION |
d254636613 | Equipping predicted segmentation with calibrated uncertainty is essential for safety-critical applications. In this work, we focus on capturing the data-inherent uncertainty (aka aleatoric uncertainty) in segmentation, typically when ambiguities exist in input images. Due to the high-dimensional output space and potential multiple modes in segmenting ambiguous images, it remains challenging to predict well-calibrated uncertainty for segmentation. To tackle this problem, we propose a novel mixture of stochastic experts (MoSE) model, where each expert network estimates a distinct mode of the aleatoric uncertainty and a gating network predicts the probabilities of an input image being segmented in those modes. This yields an efficient two-level uncertainty representation. To learn the model, we develop a Wasserstein-like loss that directly minimizes the distribution distance between the MoSE and ground truth annotations. The loss can easily integrate traditional segmentation quality measures and be efficiently optimized via constraint relaxation. We validate our method on the LIDC-IDRI dataset and a modified multimodal Cityscapes dataset. Results demonstrate that our method achieves the state-of-the-art or competitive performance on all metrics. 1 et al. The lung image database consortium (lidc) and image database resource initiative (idri): a completed reference database of lung nodules on ct scans. Medical physics, 2011. . The cramer distance as a solution to biased wasserstein gradients. arXiv preprint arXiv:1705.10743, 2017.Matan Ben-Yosef and Daphna Weinshall. Gaussian mixture generative adversarial networks for diverse datasets, and the unsupervised clustering of images. arXiv preprint arXiv:1808.10356, 2018.Yoshua Bengio, Nicholas Léonard, and Aaron Courville. Estimating or propagating gradients through stochastic neurons for conditional computation. arXiv preprint arXiv:1308.3432, 2013. decoder with atrous separable convolution for semantic image segmentation. In | Published as a conference paper at ICLR 2023 MODELING MULTIMODAL ALEATORIC UNCERTAINTY IN SEGMENTATION WITH MIXTURE OF STOCHASTIC EXPERTS |
d252596292 | Humans naturally decompose their environment into entities at the appropriate level of abstraction to act in the world. Allowing machine learning algorithms to derive this decomposition in an unsupervised way has become an important line of research. However, current methods are restricted to simulated data or require additional information in the form of motion or depth in order to successfully discover objects. In this work, we overcome this limitation by showing that reconstructing features from models trained in a self-supervised manner is a sufficient training signal for object-centric representations to arise in a fully unsupervised way. Our approach, DINOSAUR, significantly out-performs existing image-based object-centric learning models on simulated data and is the first unsupervised object-centric model that scales to real-world datasets such as COCO and PASCAL VOC. DINOSAUR is conceptually simple and shows competitive performance compared to more involved pipelines from the computer vision literature.Published as a conference paper at ICLR 2023 image datasets, which do not include depth annotations or motion cues. Following deep learning's mantra of scale, another appealing approach could be to increase the capacity of the Slot Attention architecture. However, our experiments (Sec. 4.3) suggest that scale alone is not sufficient to close the gap between synthetic and real-world datasets. We thus conjecture that the image reconstruction objective on its own does not provide sufficient inductive bias to give rise to object groupings when objects have complex appearance. But instead of relying on auxiliary external signals, we introduce an additional inductive bias by reconstructing features that have a high level of homogeneity within objects. Such features can easily be obtained via recent self-supervised learning techniques like DINO (Caron et al., 2021). We show that combining such a feature reconstruction loss with existing grouping modules such as Slot Attention leads to models that significantly out-perform other image-based object-centric methods and bridge the gap to real-world object-centric representation learning. The proposed architecture DINOSAUR (DINO and Slot Attention Using Real-world data) is conceptually simple and highly competitive with existing unsupervised segmentation and object discovery methods in computer vision. | Published as a conference paper at ICLR 2023 BRIDGING THE GAP TO REAL-WORLD OBJECT- CENTRIC LEARNING |
d69629714 | A zoo of deep nets is available these days for almost any given task, and it is increasingly unclear which net to start with when addressing a new task, or which net to use as an initialization for fine-tuning a new model. To address this issue, in this paper, we develop knowledge flow which moves 'knowledge' from multiple deep nets, referred to as teachers, to a new deep net model, called the student. The structure of the teachers and the student can differ arbitrarily and they can be trained on entirely different tasks with different output spaces too. Upon training with knowledge flow the student is independent of the teachers. We demonstrate our approach on a variety of supervised and reinforcement learning tasks, outperforming fine-tuning and other 'knowledge exchange' methods. | KNOWLEDGE FLOW: IMPROVE UPON YOUR TEACH- ERS |
d254096353 | Blind image super-resolution (Blind-SR) aims to recover a high-resolution (HR) image from its corresponding low-resolution (LR) input image with unknown degradations. Most of the existing works design an explicit degradation estimator for each degradation to guide SR. However, it is infeasible to provide concrete labels of multiple degradation combinations (e.g., blur, noise, jpeg compression) to supervise the degradation estimator training. In addition, these special designs for certain degradation, such as blur, impedes the models from being generalized to handle different degradations. To this end, it is necessary to design an implicit degradation estimator that can extract discriminative degradation representation for all degradations without relying on the supervision of degradation ground-truth. In this paper, we propose a Knowledge Distillation based Blind-SR network (KDSR). It consists of a knowledge distillation based implicit degradation estimator network (KD-IDE) and an efficient SR network. To learn the KDSR model, we first train a teacher network: KD-IDE T . It takes paired HR and LR patches as inputs and is optimized with the SR network jointly. Then, we further train a student network KD-IDE S , which only takes LR images as input and learns to extract the same implicit degradation representation (IDR) as KD-IDE T . In addition, to fully use extracted IDR, we design a simple, strong, and efficient IDR based dynamic convolution residual block (IDR-DCRB) to build an SR network. We conduct extensive experiments under classic and real-world degradation settings. The results show that KDSR achieves SOTA performance and can generalize to various degradation processes. The code is available at https://github.com/ | Published as a conference paper at ICLR 2023 KNOWLEDGE DISTILLATION BASED DEGRADATION ESTIMATION FOR BLIND SUPER-RESOLUTION |
d252439090 | Early sensory systems in the brain rapidly adapt to fluctuating input statistics, which requires recurrent communication between neurons. Mechanistically, such recurrent communication is often indirect and mediated by local interneurons. In this work, we explore the computational benefits of mediating recurrent communication via interneurons compared with direct recurrent connections. To this end, we consider two mathematically tractable recurrent linear neural networks that statistically whiten their inputs -one with direct recurrent connections and the other with interneurons that mediate recurrent communication. By analyzing the corresponding continuous synaptic dynamics and numerically simulating the networks, we show that the network with interneurons is more robust to initialization than the network with direct recurrent connections in the sense that the convergence time for the synaptic dynamics in the network with interneurons (resp. direct recurrent connections) scales logarithmically (resp. linearly) with the spectrum of their initialization. Our results suggest that interneurons are computationally useful for rapid adaptation to changing input statistics. Interestingly, the network with interneurons is an overparameterized solution of the whitening objective for the network with direct recurrent connections, so our results can be viewed as a recurrent linear neural network analogue of the implicit acceleration phenomenon observed in overparameterized feedforward linear neural networks. Carleton. Neuronal pattern separation in the olfactory bulb improves odor discrimination learning. . On feature decorrelation in self-supervised learning. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 9598-9608, 2021. Aapo Hyvärinen and Erkki Oja. Independent component analysis: algorithms and applications. Neural networks, 13(4-5):411-430, 2000. Agnan Kessy, Alex Lewin, and Korbinian Strimmer. Optimal whitening and decorrelation. The American Statistician, 72(4):309-314, 2018. Alex Krizhevsky. Learning multiple layers of features from tiny images. Master's thesis, University of Toronto, 2009. Simon B Laughlin. The role of sensory adaptation in the retina. | Published as a conference paper at ICLR 2023 INTERNEURONS ACCELERATE LEARNING DYNAMICS IN RECURRENT NEURAL NETWORKS FOR STATISTICAL ADAPTATION |
d13583585 | The past several years have seen remarkable progress in generative models which produce convincing samples of images and other modalities. A shared component of many powerful generative models is a decoder network, a parametric deep neural net that defines a generative distribution. Examples include variational autoencoders, generative adversarial networks, and generative moment matching networks. Unfortunately, it can be difficult to quantify the performance of these models because of the intractability of log-likelihood estimation, and inspecting samples can be misleading. We propose to use Annealed Importance Sampling for evaluating log-likelihoods for decoder-based models and validate its accuracy using bidirectional Monte Carlo. The evaluation code is provided at https:// github.com/tonywu95/eval_gen. Using this technique, we analyze the performance of decoder-based models, the effectiveness of existing log-likelihood estimators, the degree of overfitting, and the degree to which these models miss important modes of the data distribution. | Published as a conference paper at ICLR 2017 ON THE QUANTITATIVE ANALYSIS OF DECODER- BASED GENERATIVE MODELS |
d203836194 | Convolutional networks are not aware of an object's geometric variations, which leads to inefficient utilization of model and data capacity.To overcome this issue, recent works on deformation modeling seek to spatially reconfigure the data towards a common arrangement such that semantic recognition suffers less from deformation.This is typically done by augmenting static operators with learned free-form sampling grids in the image space, dynamically tuned to the data and task for adapting the receptive field.Yet adapting the receptive field does not quite reach the actual goal -what really matters to the network is the effective receptive field (ERF), which reflects how much each pixel contributes.It is thus natural to design other approaches to adapt the ERF directly during runtime.In this work, we instantiate one possible solution as Deformable Kernels (DKs), a family of novel and generic convolutional operators for handling object deformations by directly adapting the ERF while leaving the receptive field untouched.At the heart of our method is the ability to resample the original kernel space towards recovering the deformation of objects.This approach is justified with theoretical insights that the ERF is strictly determined by data sampling locations and kernel values.We implement DKs as generic drop-in replacements of rigid kernels and conduct a series of empirical studies whose results conform with our theories.Over several tasks and standard base models, our approach compares favorably against prior works that adapt during runtime.In addition, further experiments suggest a working mechanism orthogonal and complementary to previous works. | DEFORMABLE KERNELS: ADAPTING EFFECTIVE RE-CEPTIVE FIELDS FOR OBJECT DEFORMATION |
d235254582 | Understanding how large neural networks avoid memorizing training data is key to explaining their high generalization performance. To examine the structure of when and where memorization occurs in a deep network, we use a recently developed replica-based mean field theoretic geometric analysis method. We find that all layers preferentially learn from examples which share features, and link this behavior to generalization performance. Memorization predominately occurs in the deeper layers, due to decreasing object manifolds' radius and dimension, whereas early layers are minimally affected. This predicts that generalization can be restored by reverting the final few layer weights to earlier epochs before significant memorization occurred, which is confirmed by the experiments. Additionally, by studying generalization under different model sizes, we reveal the connection between the double descent phenomenon and the underlying model geometry. Finally, analytical analysis shows that networks avoid memorization early in training because close to initialization, the gradient contribution from permuted examples are small. These findings provide quantitative evidence for the structure of memorization across layers of a deep neural network, the drivers for such structure, and its connection to manifold geometric properties.Recent work have shown that a combination of architecture and stochastic gradient descent implicitly bias the training dynamics towards generalizable solutions(Hardt et al., 2016;Soudry et al., 2018;Brutzkus et al., 2017;Li and Liang, 2018; Saxe et al., 2013; Lampinen and Ganguli, 2018). However, these claims study either linear networks or two-layer non-linear networks. For deep neural networks, open questions remain on the structure of memorization, such as where and when in the layers of the network is memorization occurring (e.g. evenly across all layers, gradually increasing with depth, or concentrated in early or late layers), and what are the drivers of this structure.Analytical tools for linear networks such as eigenvalue decomposition cannot be directly applied to non-linear networks, so here we employ a recently developed geometric probe(Chung et al., 2018;Stephenson et al., 2019), based on replica mean field theory from statistical physics, to analyze the training dynamics and resulting structure of memorization. The probe measures not just the layer capacity, but also geometric properties of the object manifolds, explicitly linked by the theory.We find that deep neural networks ignore randomly labeled data in the early layers and epochs, instead learning generalizing features. Memorization occurs abruptly with depth in the final layers, caused by Published as a conference paper at ICLR 2021 decreasing manifold radius and dimension, whereas early layers are minimally affected. Notably, this structure does not arise due to gradients vanishing with depth. Instead, analytical analysis show that near initialization, the gradients from noise examples contribute minimally to the total gradient, and that networks are able to ignore noise due to the existence of 'shared features' consisting of linear features shared by objects of the same class. Of practical consequence, generalization can then be re-gained by rolling back the parameters of the final layers of the network to an earlier epoch before the structural signatures of memorization occur.Moreover, the 'double descent' phenomenon, where a model's generalization performance initially decreases with model size before increasing, is linked to the non-monotonic dimensionality expansion of the object manifolds, as measured by the geometric probe. The manifold dimensionality also undergoes double descent, whereas other geometric measures, such as radius and center correlation are monotonic with model size.Our analysis reveals the structure of memorization in deep networks, and demonstrate the importance of measuring manifold geometric properties in tracing the effect of learning on neural networks.RELATED WORKBy demonstrating that deep neural networks can easily fit random labels with standard training procedures,Zhang et al. (2016)showed that standard regularization strategies are not enough to prevent memorization. Since then, several works have explored the problem experimentally. NotablyArpit et al. (2017)examined the behavior of the network as a single unit when trained on random data, and observed that the training dynamics were qualitatively different when the network was trained on real data vs. noise. They observed experimentally that networks learn from 'simple' patterns in the data first when trained with gradient descent, but do not give a rigorous explanation of this effect. | ON THE GEOMETRY OF GENERALIZATION AND MEMO- RIZATION IN DEEP NEURAL NETWORKS |
d256389374 | Heteroscedastic classifiers, which learn a multivariate Gaussian distribution over prediction logits, have been shown to perform well on image classification problems with hundreds to thousands of classes. However, compared to standard classifiers, they introduce extra parameters that scale linearly with the number of classes. This makes them infeasible to apply to larger-scale problems. In addition heteroscedastic classifiers introduce a critical temperature hyperparameter which must be tuned. We propose HET-XL, a heteroscedastic classifier whose parameter count when compared to a standard classifier scales independently of the number of classes. In our large-scale settings, we show that we can remove the need to tune the temperature hyperparameter, by directly learning it on the training data. On large image classification datasets with up to 4B images and 30k classes our method requires 14× fewer additional parameters, does not require tuning the temperature on a held-out set and performs consistently better than the baseline heteroscedastic classifier. HET-XL improves ImageNet 0-shot classification in a multimodal contrastive learning setup which can be viewed as a 3.5 billion class classification problem. *Equal contribution. | Published as a conference paper at ICLR 2023 MASSIVELY SCALING HETEROSCEDASTIC CLASSIFIERS |
d52962991 | When a black-box classifier processes an input to render a prediction, which input features are relevant and why? We propose to answer this question by efficiently marginalizing over the universe of plausible alternative values for a subset of features by conditioning a generative model of the input distribution on the remaining features. In contrast with recent approaches that compute alternative feature values ad-hoc-generating counterfactual inputs far from the natural data distributionour model-agnostic method produces realistic explanations, generating plausible inputs that either preserve or alter the classification confidence. When applied to image classification, our method produces more compact and relevant per-feature saliency assignment, with fewer artifacts compared to previous methods. | EXPLAINING IMAGE CLASSIFIERS BY COUNTERFAC- TUAL GENERATION |
d38796293 | Large-scale distributed training requires significant communication bandwidth for gradient exchange that limits the scalability of multi-node training, and requires expensive high-bandwidth network infrastructure. The situation gets even worse with distributed training on mobile devices (federated learning), which suffers from higher latency, lower throughput, and intermittent poor connections. In this paper, we find 99.9% of the gradient exchange in distributed SGD are redundant, and propose Deep Gradient Compression (DGC) to greatly reduce the communication bandwidth. To preserve accuracy during this compression, DGC employs four methods: momentum correction, local gradient clipping, momentum factor masking, and warm-up training. We have applied Deep Gradient Compression to image classification, speech recognition, and language modeling with multiple datasets including Cifar10, ImageNet, Penn Treebank, and Librispeech Corpus. On these scenarios, Deep Gradient Compression achieves a gradient compression ratio from 270× to 600× without losing accuracy, cutting the gradient size of ResNet-50 from 97MB to 0.35MB, and for DeepSpeech from 488MB to 0.74MB. Deep gradient compression enables large-scale distributed training on inexpensive commodity 1Gbps Ethernet and facilitates distributed training on mobile. | Published as a conference paper at ICLR 2018 DEEP GRADIENT COMPRESSION: REDUCING THE COMMUNICATION BANDWIDTH FOR DISTRIBUTED TRAINING |
d257279837 | Spectral graph neural networks (GNNs) learn graph representations via spectraldomain graph convolutions. However, most existing spectral graph filters are scalar-to-scalar functions, i.e., mapping a single eigenvalue to a single filtered value, thus ignoring the global pattern of the spectrum. Furthermore, these filters are often constructed based on some fixed-order polynomials, which have limited expressiveness and flexibility. To tackle these issues, we introduce Specformer, which effectively encodes the set of all eigenvalues and performs self-attention in the spectral domain, leading to a learnable set-to-set spectral filter. We also design a decoder with learnable bases to enable non-local graph convolution. Importantly, Specformer is equivariant to permutation. By stacking multiple Specformer layers, one can build a powerful spectral GNN. On synthetic datasets, we show that our Specformer can better recover ground-truth spectral filters than other spectral GNNs. Extensive experiments of both node-level and graph-level tasks on real-world graph datasets show that our Specformer outperforms state-ofthe-art GNNs and learns meaningful spectrum patterns. | SPECFORMER: SPECTRAL GRAPH NEURAL NET- WORKS MEET TRANSFORMERS |
d237490346 | Unsupervised domain adaptation (UDA) aims to transfer knowledge learned from a labeled source domain to a different unlabeled target domain. Most existing UDA methods focus on learning domain-invariant feature representation, either from the domain level or category level, using convolution neural networks (CNNs)-based frameworks. One fundamental problem for the category level based UDA is the production of pseudo labels for samples in target domain, which are usually too noisy for accurate domain alignment, inevitably compromising the UDA performance. With the success of Transformer in various tasks, we find that the cross-attention in Transformer is robust to the noisy input pairs for better feature alignment, thus in this paper Transformer is adopted for the challenging UDA task. Specifically, to generate accurate input pairs, we design a two-way center-aware labeling algorithm to produce pseudo labels for target samples. Along with the pseudo labels, a weight-sharing triple-branch transformer framework is proposed to apply self-attention and cross-attention for source/target feature learning and source-target domain alignment, respectively. Such design explicitly enforces the framework to learn discriminative domain-specific and domain-invariant representations simultaneously. The proposed method is dubbed CDTrans (cross-domain transformer), and it provides one of the first attempts to solve UDA tasks with a pure transformer solution. Experiments show that our proposed method achieves the best performance on public UDA datasets, e.g. VisDA-2017 and DomainNet. Code and models are available at https: //github.com/CDTrans/CDTrans. * These authors contributed equally to this work. | CDTRANS: CROSS-DOMAIN TRANSFORMER FOR UN- SUPERVISED DOMAIN ADAPTATION |
d243861189 | Graph Neural Networks (GNNs) have achieved state-of-the-art performance in node classification, regression, and recommendation tasks. GNNs work well when rich and high-quality connections are available. However, their effectiveness is often jeopardized in many real-world graphs in which node degrees have power-law distributions. The extreme case of this situation, where a node may have no neighbors, is called Strict Cold Start (SCS). SCS forces the prediction to rely completely on the node's own features. We propose Cold Brew, a teacher-student distillation approach to address the SCS and noisy-neighbor challenges for GNNs. We also introduce feature contribution ratio (FCR), a metric to quantify the behavior of inductive GNNs to solve SCS. We experimentally show that FCR disentangles the contributions of different graph data components and helps select the best architecture for SCS generalization. We further demonstrate the superior performance of Cold Brew on several public benchmark and proprietary e-commerce datasets, where many nodes have either very few or noisy connections. Our source code is available at https: //github.com/amazon-research/gnn-tail-generalization. | Published as a conference paper at ICLR 2022 COLD BREW: DISTILLING GRAPH NODE REPRESEN- TATIONS WITH INCOMPLETE OR MISSING NEIGHBOR- HOODS |
d3457087 | One of the distinguishing aspects of human language is its compositionality, which allows us to describe complex environments with limited vocabulary. Previously, it has been shown that neural network agents can learn to communicate in a highly structured, possibly compositional language based on disentangled input (e.g. handengineered features). Humans, however, do not learn to communicate based on well-summarized features. In this work, we train neural agents to simultaneously develop visual perception from raw image pixels, and learn to communicate with a sequence of discrete symbols. The agents play an image description game where the image contains factors such as colors and shapes. We train the agents using the obverter technique where an agent introspects to generate messages that maximize its own understanding. Through qualitative analysis, visualization and a zero-shot test, we show that the agents can develop, out of raw image pixels, a language with compositional properties, given a proper pressure from the environment. * Work done as an intern at DeepMind. | Published as a conference paper at ICLR 2018 COMPOSITIONAL OBVERTER COMMUNICATION LEARNING FROM RAW VISUAL INPUT |
d208527260 | Traditional compression methods including network pruning, quantization, low rank factorization and knowledge distillation all assume that network architectures and parameters are one-to-one mapped. In this work, we propose a new perspective on network compression, i.e., network parameters can be disentangled from the architectures. From this viewpoint, we present the Neural Epitome Search (NES), a new neural network compression approach that learns to find compact yet expressive epitomes for weight parameters of a specified network architecture end-to-end. The complete network to compress can be generated from the learned epitome via a novel transformation method that adaptively transforms the epitomes to match weight shapes of the given architecture. Compared with existing compression methods, NES allows the weight tensors to be independent of the architecture design and hence can achieve a good trade-off between model compression rate and performance given a specific model size constraint. Experiments demonstrate that, on ImageNet, when taking MobileNetV2 as backbone, our approach improves the full-model baseline by 1.47% in top-1 accuracy with 25% MAdd reduction, and with the same compression ratio, improves AutoML for Model Compression (AMC) by 2.5% in top-1 accuracy. Moreover, taking EfficientNet-B0 as baseline, our NES yields an improvement of 1.2% but has 10% less MAdd. In particular, our method achieves a new state-of-the-art results of 77.5% under mobile settings (<350M MAdd). Code will be made publicly available. | NEURAL EPITOME SEARCH FOR ARCHITECTURE- AGNOSTIC NETWORK COMPRESSION |
d211133221 | Robustness verification that aims to formally certify the prediction behavior of neural networks has become an important tool for understanding model behavior and obtaining safety guarantees. However, previous methods can usually only handle neural networks with relatively simple architectures. In this paper, we consider the robustness verification problem for Transformers. Transformers have complex self-attention layers that pose many challenges for verification, including cross-nonlinearity and cross-position dependency, which have not been discussed in previous works. We resolve these challenges and develop the first robustness verification algorithm for Transformers. The certified robustness bounds computed by our method are significantly tighter than those by naive Interval Bound Propagation. These bounds also shed light on interpreting Transformers as they consistently reflect the importance of different words in sentiment analysis.Li et al., 2019a). For frames under perturbation in the input sequence, we aim to compute a lower bound such that when these frames are perturbed within p -balls centered at the original frames respectively and with a radius of , the model prediction is certified to be unchanged. To compute such bounds efficiently, we adopt the linear-relaxation framework (Weng et al., 2018;Zhang et al., 2018) -we recursively propagate and compute linear lower and upper bounds for each neuron w.r.t the input within the perturbation space S.We resolve several particular challenges in verifying Transformers. First, Transformers with selfattention layers have a complicated architecture. Unlike simpler networks, they cannot be written as multiple layers of affine transformations or element-wise activation functions. Therefore, we need to propagate linear bounds differently for self-attention layers. Second, dot products, softmax, and weighted summation in self-attention layers involve multiplication or division of two variables both under perturbation, namely cross-nonlinearity, which is not present in feed-forward networks.Ko et al. (2019)proposed a gradient descent based approach to find linear bounds, however it is inefficient and poses a computational challenge for Transformer verification since self-attention is the core of Transformers. In contrast, we derive closed-form linear bounds that can be computed in O(1) complexity. Third, in the computation of self-attention, output neurons in each position depend on all input neurons from different positions (namely cross-position dependency), unlike the case in recurrent neural networks where outputs depend on only the hidden features from the previous position and the current input. Previous works(Zhang et al., 2018;Weng et al., 2018;Ko et al., 2019)have to track all such dependency and thus is costly in time and memory. To tackle this, we introduce an efficient bound propagating process in a forward manner specially for self-attention layers, enabling the tighter backward bounding process for other layers to utilize bounds computed by the forward process. In this way, we avoid cross-position dependency in the backward process which is relatively slower but produces tighter bounds. Combined with the forward process, the complexity of the backward process is reduced by O(n) for input length n, while the computed bounds remain comparably tight. Our contributions are summarized below: | Published as a conference paper at ICLR 2020 ROBUSTNESS VERIFICATION FOR TRANSFORMERS |
d224726548 | Data augmentation has been demonstrated as an effective strategy for improving model generalization and data efficiency. However, due to the discrete nature of natural language, designing label-preserving transformations for text data tends to be more challenging. In this paper, we propose a novel data augmentation framework dubbed CoDA, which synthesizes diverse and informative augmented examples by integrating multiple transformations organically. Moreover, a contrastive regularization objective is introduced to capture the global relationship among all the data samples. A momentum encoder along with a memory bank is further leveraged to better estimate the contrastive loss. To verify the effectiveness of the proposed framework, we apply CoDA to Transformer-based models on a wide range of natural language understanding tasks. On the GLUE benchmark, CoDA gives rise to an average improvement of 2.2% while applied to the Roberta-large model. More importantly, it consistently exhibits stronger results relative to several competitive data augmentation and adversarial training baselines (including the low-resource settings). Extensive experiments show that the proposed contrastive objective can be flexibly combined with various data augmentation approaches to further boost their performance, highlighting the wide applicability of the CoDA framework. * Work was done during an internship at Microsoft Dynamics 365 AI. 1 To name a few, BERT (Devlin et al., 2019): 340M parameters, T5 (Raffel et al., 2019): 11B parameters, GPT-3 (Brown et al., 2020): 175B parameters. | CODA: CONTRAST-ENHANCED AND DIVERSITY- PROMOTING DATA AUGMENTATION FOR NATURAL LANGUAGE UNDERSTANDING |
d211107043 | Deep representation learning has become one of the most widely adopted approaches for visual search, recommendation, and identification. Retrieval of such representations from a large database is however computationally challenging. Approximate methods based on learning compact representations, have been widely explored for this problem, such as locality sensitive hashing, product quantization, and PCA. In this work, in contrast to learning compact representations, we propose to learn high dimensional and sparse representations that have similar representational capacity as dense embeddings while being more efficient due to sparse matrix multiplication operations which can be much faster than dense multiplication. Following the key insight that the number of operations decreases quadratically with the sparsity of embeddings provided the non-zero entries are distributed uniformly across dimensions, we propose a novel approach to learn such distributed sparse embeddings via the use of a carefully constructed regularization function that directly minimizes a continuous relaxation of the number of floating-point operations (FLOPs) incurred during retrieval. Our experiments show that our approach is competitive to the other baselines and yields a similar or better speed-vs-accuracy tradeoff on practical datasets 1 . | Published as a conference paper at ICLR 2020 MINIMIZING FLOPS TO LEARN EFFICIENT SPARSE REPRESENTATIONS |
d246652175 | Owing much to the revolution of information technology, the recent progress of deep learning benefits incredibly from the vastly enhanced access to data available in various digital formats. However, in certain scenarios, people may not want their data being used for training commercial models and thus studied how to attack the learnability of deep learning models. Previous works on learnability attack only consider the goal of preventing unauthorized exploitation on the specific dataset but not the process of restoring the learnability for authorized cases. To tackle this issue, this paper introduces and investigates a new concept called "learnability lock" for controlling the model's learnability on a specific dataset with a special key. In particular, we propose adversarial invertible transformation, that can be viewed as a mapping from image to image, to slightly modify data samples so that they become "unlearnable" by machine learning models with negligible loss of visual features. Meanwhile, one can unlock the learnability of the dataset and train models normally using the corresponding key. The proposed learnability lock leverages class-wise perturbation that applies a universal transformation function on data samples of the same label. This ensures that the learnability can be easily restored with a simple inverse transformation while remaining difficult to be detected or reverse-engineered. We empirically demonstrate the success and practicability of our method on visual classification tasks. | Published as a conference paper at ICLR 2022 LEARNABILITY LOCK: AUTHORIZED LEARNABIL- ITY CONTROL THROUGH ADVERSARIAL INVERTIBLE TRANSFORMATIONS |
d247996596 | Recent text-to-image models have achieved impressive results. However, since they require large-scale datasets of text-image pairs, it is impractical to train them on new domains where data is scarce or not labeled. In this work, we propose using large-scale retrieval methods, in particular, efficient k-Nearest-Neighbors (kNN), which offers novel capabilities: (1) training a substantially small and efficient text-to-image diffusion model without any text, (2) generating out-ofdistribution images by simply swapping the retrieval database at inference time, and (3) performing text-driven local semantic manipulations while preserving object identity. To demonstrate the robustness of our method, we apply our kNN approach on two state-of-the-art diffusion backbones, and show results on several different datasets. As evaluated by human studies and automatic metrics, our method achieves state-of-the-art results compared to existing approaches that train text-to-image generation models using images only (without paired text data). | KNN-Diffusion: Image Generation via Large-Scale Retrieval Figure 1: (a) Samples of stickers generated from text inputs, (b) Semantic text-guided manipulations applied to the "Original" image without using edit masks. In both cases, our model was trained without any text data |
d17267607 | One of the challenges in modeling cognitive events from electroencephalogram (EEG) data is finding representations that are invariant to inter-and intra-subject differences, as well as to inherent noise associated with EEG data collection. Herein, we propose a novel approach for learning such representations from multichannel EEG time-series, and demonstrate its advantages in the context of mental load classification task. First, we transform EEG activities into a sequence of topology-preserving multi-spectral images, as opposed to standard EEG analysis techniques that ignore such spatial information. Next, we train a deep recurrentconvolutional network inspired by state-of-the-art video classification techniques to learn robust representations from the sequence of images. The proposed approach is designed to preserve the spatial, spectral, and temporal structure of EEG which leads to finding features that are less sensitive to variations and distortions within each dimension. Empirical evaluation on the cognitive load classification task demonstrated significant improvements in classification accuracy over current state-of-the-art approaches in this field. | LEARNING REPRESENTATIONS FROM EEG WITH DEEP RECURRENT-CONVOLUTIONAL NEURAL NETWORKS |
d13748235 | Current reinforcement learning (RL) methods can successfully learn single tasks, but often generalize poorly to modest perturbations in task domain or training procedure. In this work we present a decoupled learning strategy for RL that creates a shared representation space where knowledge can be robustly transferred. We separate learning the task representation, the forward dynamics, the inverse dynamics and the reward function of the domain, and show that this decoupling improves performance within task, transfers well to changes in dynamics and reward, and can be effectively used for online planning. Empirical results show good performance in both continuous and discrete RL domains. | Decoupling Dynamics and Reward for Transfer Learning |
d257505065 | AutoML has demonstrated remarkable success in finding an effective neural architecture for a given machine learning task defined by a specific dataset and an evaluation metric. However, most present AutoML techniques consider each task independently from scratch, which requires exploring many architectures, leading to high computational costs. Here we propose AUTOTRANSFER, an AutoML solution that improves search efficiency by transferring the prior architectural design knowledge to the novel task of interest. Our key innovation includes a task-model bank that captures the model performance over a diverse set of GNN architectures and tasks, and a computationally efficient task embedding that can accurately measure the similarity among different tasks. Based on the task-model bank and the task embeddings, we estimate the design priors of desirable models of the novel task, by aggregating a similarity-weighted sum of the top-K design distributions on tasks that are similar to the task of interest. The computed design priors can be used with any AutoML search algorithm. We evaluate AUTOTRANSFER on six datasets in the graph machine learning domain. Experiments demonstrate that (i) our proposed task embedding can be computed efficiently, and that tasks with similar embeddings have similar best-performing architectures; (ii) AUTOTRANSFER significantly improves search efficiency with the transferred design priors, reducing the number of explored architectures by an order of magnitude. Finally, we release GNN-BANK-101, a large-scale dataset of detailed GNN training information of 120,000 task-model combinations to facilitate and inspire future research. | AUTOTRANSFER: AUTOML WITH KNOWLEDGE TRANSFER -AN APPLICATION TO GRAPH NEURAL NETWORKS |
d492235 | This paper is focused on studying the view-manifold structure in the feature spaces implied by the different layers of Convolutional Neural Networks (CNN). There are several questions that this paper aims to answer: Does the learned CNN representation achieve viewpoint invariance? How does it achieve viewpoint invariance? Is it achieved by collapsing the view manifolds, or separating them while preserving them? At which layer is view invariance achieved? How can the structure of the view manifold at each layer of a deep convolutional neural network be quantified experimentally? How does fine-tuning of a pre-trained CNN on a multi-view dataset affect the representation at each layer of the network? In order to answer these questions we propose a methodology to quantify the deformation and degeneracy of view manifolds in CNN layers. We apply this methodology and report interesting results in this paper that answer the aforementioned questions. | DIGGING DEEP INTO THE LAYERS OF CNNS: IN SEARCH OF HOW CNNS ACHIEVE VIEW INVARIANCE |
d256662685 | Open-set Recognition (OSR) aims to identify test samples whose classes are not seen during the training process. Recently, Unified Open-set Recognition (UOSR) has been proposed to reject not only unknown samples but also known but wrongly classified samples, which tends to be more practical in real-world applications. In this paper, we deeply analyze the UOSR task under different training and evaluation settings to shed light on this promising research direction. For this purpose, we first evaluate the UOSR performance of several OSR methods and show a significant finding that the UOSR performance consistently surpasses the OSR performance by a large margin for the same method. We show that the reason lies in the known but wrongly classified samples, as their uncertainty distribution is extremely close to unknown samples rather than known and correctly classified samples. Second, we analyze how the two training settings of OSR (i.e., pre-training and outlier exposure) influence the UOSR. We find although they are both beneficial for distinguishing known and correctly classified samples from unknown samples, pre-training is also helpful for identifying known but wrongly classified samples while outlier exposure is not. In addition to different training settings, we also formulate a new evaluation setting for UOSR which is called few-shot UOSR, where only one or five samples per unknown class are available during evaluation to help identify unknown samples. We propose FS-KNNS for the few-shot UOSR to achieve state-of-the-art performance under all settings. * Equal contribution. Code: https://github.com/Jun-CEN/Unified_Open_Set_Recognition. Published as a conference paper at ICLR 2023TOWARDS UNIFIED OPEN-SET RECOGNITIONIn this section, we first formalize the UOSR problem and then discuss the relation between UOSR and other uncertainty-related tasks.Unified Open-set Recognition. Suppose the training dataset is D train = {(x i , y i )} N i=1 ⊂ X × C, where X refers to the input space, e.g., images or videos, and C refers to the InD sets. In closed-set Terrance DeVries and Graham W Taylor. Learning confidence for out-of-distribution detection in neural networks. arXiv preprint arXiv:1802.04865, 2018. unified benchmark for the unknown detection capability of deep neural networks. arXiv preprint arXiv:2112.00337, 2021. | THE DEVIL IS IN THE WRONGLY-CLASSIFIED SAM- PLES: TOWARDS UNIFIED OPEN-SET RECOGNITION |
d232013680 | Neural Architecture Search (NAS) has been explosively studied to automate the discovery of top-performer neural networks. Current works require heavy training of supernet or intensive architecture evaluations, thus suffering from heavy resource consumption and often incurring search bias due to truncated training or approximations. Can we select the best neural architectures without involving any training and eliminate a drastic portion of the search cost? We provide an affirmative answer, by proposing a novel framework called training-free neural architecture search (TE-NAS). TE-NAS ranks architectures by analyzing the spectrum of the neural tangent kernel (NTK) and the number of linear regions in the input space. Both are motivated by recent theory advances in deep networks and can be computed without any training and any label. We show that: (1) these two measurements imply the trainability and expressivity of a neural network; (2) they strongly correlate with the network's test accuracy. Further on, we design a pruning-based NAS mechanism to achieve a more flexible and superior trade-off between the trainability and expressivity during the search. In NAS-Bench-201 and DARTS search spaces, TE-NAS completes high-quality search but only costs 0.5 and 4 GPU hours with one 1080Ti on CIFAR-10 and ImageNet, respectively. We hope our work inspires more attempts in bridging the theoretical findings of deep networks and practical impacts in real NAS applications. Code is available at: https://github.com/VITA-Group/TENAS. | Published as a conference paper at ICLR 2021 NEURAL ARCHITECTURE SEARCH ON IMAGENET IN FOUR GPU HOURS: A THEORETICALLY INSPIRED PERSPECTIVE |
d233181741 | Recently, a variety of probing tasks are proposed to discover linguistic properties learned in contextualized word embeddings. Many of these works implicitly assume these embeddings lay in certain metric spaces, typically the Euclidean space. This work considers a family of geometrically special spaces, the hyperbolic spaces, that exhibit better inductive biases for hierarchical structures and may better reveal linguistic hierarchies encoded in contextualized representations. We introduce a Poincaré probe, a structural probe projecting these embeddings into a Poincaré subspace with explicitly defined hierarchies. We focus on two probing objectives: (a) dependency trees where the hierarchy is defined as headdependent structures; (b) lexical sentiments where the hierarchy is defined as the polarity of words (positivity and negativity). We argue that a key desideratum of a probe is its sensitivity to the existence of linguistic structures. We apply our probes on BERT, a typical contextualized embedding model. In a syntactic subspace, our probe better recovers tree structures than Euclidean probes, revealing the possibility that the geometry of BERT syntax may not necessarily be Euclidean. In a sentiment subspace, we reveal two possible meta-embeddings for positive and negative sentiments and show how lexically-controlled contextualization would change the geometric localization of embeddings. We demonstrate the findings with our Poincaré probe via extensive experiments and visualization 1 | Published as a conference paper at ICLR 2021 PROBING BERT IN HYPERBOLIC SPACES |
d35742668 | Most machine learning classifiers, including deep neural networks, are vulnerable to adversarial examples. Such inputs are typically generated by adding small but purposeful modifications that lead to incorrect outputs while imperceptible to human eyes. The goal of this paper is not to introduce a single method, but to make theoretical steps towards fully understanding adversarial examples. By using concepts from topology, our theoretical analysis brings forth the key reasons why an adversarial example can fool a classifier (f 1 ) and adds its oracle (f 2 , like human eyes) in such analysis. By investigating the topological relationship between two (pseudo)metric spaces corresponding to predictor f 1 and oracle f 2 , we develop necessary and sufficient conditions that can determine if f 1 is always robust (strongrobust) against adversarial examples according to f 2 . Interestingly our theorems indicate that just one unnecessary feature can make f 1 not strong-robust, and the right feature representation learning is the key to getting a classifier that is both accurate and strong-robust. arXiv:1505.06556, 2015.Wei Liu and Sanjay Chawla. Mining adversarial patterns via regularized loss minimization. Machine learning, 81(1):69-83, 2010. | Workshop track -ICLR 2017 A THEORETICAL FRAMEWORK FOR ROBUSTNESS OF (DEEP) CLASSIFIERS AGAINST ADVERSARIAL EXAMPLES |
d238408313 | A more realistic object detection paradigm, Open-World Object Detection, has arised increasing research interests in the community recently. A qualified openworld object detector can not only identify objects of known categories, but also discover unknown objects, and incrementally learn to categorize them when their annotations progressively arrive. Previous works rely on independent modules to recognize unknown categories and perform incremental learning, respectively. In this paper, we provide a unified perspective: Semantic Topology. During the life-long learning of an open-world object detector, all object instances from the same category are assigned to their corresponding pre-defined node in the semantic topology, including the 'unknown' category. This constraint builds up discriminative feature representations and consistent relationships among objects, thus enabling the detector to distinguish unknown objects out of the known categories, as well as making learned features of known objects undistorted when learning new categories incrementally. Extensive experiments demonstrate that semantic topology, either randomly-generated or derived from a well-trained language model, could outperform the current state-of-the-art open-world object detectors by a large margin, e.g., the absolute open-set error (the number of unknown instances that are wrongly labeled as known) is reduced from 7832 to 2546, exhibiting the inherent superiority of semantic topology on open-world object detection. | Published as a conference paper at ICLR 2022 OBJECTS IN SEMANTIC TOPOLOGY |
d210473577 | A central question of representation learning asks under which conditions it is possible to reconstruct the true latent variables of an arbitrarily complex generative process. Recent breakthrough work by Khemakhem et al. (2019) on nonlinear ICA has answered this question for a broad class of conditional generative processes. We extend this important result in a direction relevant for application to real-world data. First, we generalize the theory to the case of unknown intrinsic problem dimension and prove that in some special (but not very restrictive) cases, informative latent variables will be automatically separated from noise by an estimating model. Furthermore, the recovered informative latent variables will be in one-to-one correspondence with the true latent variables of the generating process, up to a trivial component-wise transformation. Second, we introduce a modification of the RealNVP invertible neural network architecture(Dinh et al., 2016)which is particularly suitable for this type of problem: the General Incompressibleflow Network (GIN). Experiments on artificial data and EMNIST demonstrate that theoretical predictions are indeed verified in practice. In particular, we provide a detailed set of exactly 22 informative latent variables extracted from EMNIST.Published as a conference paper at ICLR 2020 Welling, 2013) used in that work. For us, major advantages of invertible architectures over VAEs are the ability to specify volume preservation and directly optimize the likelihood, and freedom from the requirement to specify the dimension of the model's latent space. An INN always has a latent space of the same dimension as the data. In addition, the forward and backward models share parameters, saving the effort of learning separate models for each direction.In summary, our work makes the following contributions:• We extend the theory of nonlinear ICA to allow for unknown intrinsic problem dimension.Doing so, we find that this dimension can be discovered and a one-to-one correspondence between generating and estimated latent variables established. • We propose as an implementation an invertible neural network obtained by modifying the RealNVP architecture. We call our new architecture GIN: the General Incompressible-flow Network. • We demonstrate the viability of the model on artificial data and the EMNIST dataset. We extract 22 meaningful variables from EMNIST, encoding both global and local features.Aapo Hyvärinen and Hiroshi Morioka. Unsupervised feature extraction by time-contrastive learning and nonlinear ICA. Composing graphical models with neural networks for structured representations and fast inference. In Advances in neural information processing systems, pp. 2946-2954, 2016. | Published as a conference paper at ICLR 2020 DISENTANGLEMENT BY NONLINEAR ICA WITH GENERAL INCOMPRESSIBLE-FLOW NETWORKS (GIN) |
d13056261 | We investigate deep generative models that can exchange multiple modalities bidirectionally, e.g., generating images from corresponding texts and vice versa. Recently, some studies handle multiple modalities on deep generative models, such as variational autoencoders (VAEs). However, these models typically assume that modalities are forced to have a conditioned relation, i.e., we can only generate modalities in one direction. To achieve our objective, we should extract a joint representation that captures high-level concepts among all modalities and through which we can exchange them bi-directionally. As described herein, we propose a joint multimodal variational autoencoder (JMVAE), in which all modalities are independently conditioned on joint representation. In other words, it models a joint distribution of modalities. Furthermore, to be able to generate missing modalities from the remaining modalities properly, we develop an additional method, JMVAE-kl, that is trained by reducing the divergence between JMVAE's encoder and prepared networks of respective modalities. Our experiments show that our proposed method can obtain appropriate joint representation from multiple modalities and that it can generate and reconstruct them more properly than conventional VAEs. We further demonstrate that JMVAE can generate multiple modalities bidirectionally. | JOINT MULTIMODAL LEARNING WITH DEEP GENERA- TIVE MODELS |
d16342357 | It is an important task to learn a representation for images which has low dimension and preserve the valuable information in original space. At the perspective of manifold, this is conduct by using a series of local invariant mapping. Inspired by the recent successes of deep architectures, we propose a local invariant deep nonlinear mapping algorithm, called graph regularized auto-encoder (GAE). The local invariant is achieved using a graph regularizer, which preserves the local Euclidean property from original space to the representation space, while the deep nonlinear mapping is based on an unsupervised trained deep auto-encoder. This provides an alternative option to current deep representation learning techniques with its competitive performance compared to these methods, as well as existing local invariant methods. | Image Representation Learning Using Graph Regularized Auto-Encoders |
d233739662 | We present Mixture of Contrastive Experts (MiCE), a unified probabilistic clustering framework that simultaneously exploits the discriminative representations learned by contrastive learning and the semantic structures captured by a latent mixture model. Motivated by the mixture of experts, MiCE employs a gating function to partition an unlabeled dataset into subsets according to the latent semantics and multiple experts to discriminate distinct subsets of instances assigned to them in a contrastive learning manner. To solve the nontrivial inference and learning problems caused by the latent variables, we further develop a scalable variant of the Expectation-Maximization (EM) algorithm for MiCE and provide proof of the convergence. Empirically, we evaluate the clustering performance of MiCE on four widely adopted natural image datasets. MiCE achieves significantly better results 1 than various previous methods and a strong contrastive learning baseline. | MICE: MIXTURE OF CONTRASTIVE EXPERTS FOR UN- SUPERVISED IMAGE CLUSTERING |
d53218829 | In open-domain dialogue intelligent agents should exhibit the use of knowledge, however there are few convincing demonstrations of this to date. The most popular sequence to sequence models typically "generate and hope" generic utterances that can be memorized in the weights of the model when mapping from input utterance(s) to output, rather than employing recalled knowledge as context. Use of knowledge has so far proved difficult, in part because of the lack of a supervised learning benchmark task which exhibits knowledgeable open dialogue with clear grounding. To that end we collect and release a large dataset with conversations directly grounded with knowledge retrieved from Wikipedia. We then design architectures capable of retrieving knowledge, reading and conditioning on it, and finally generating natural responses. Our best performing dialogue models are able to conduct knowledgeable discussions on open-domain topics as evaluated by automatic metrics and human evaluations, while our new benchmark allows for measuring further improvements in this important research direction. * Joint first authors. | OF WIKIPEDIA: KNOWLEDGE-POWERED CONVERSATIONAL AGENTS |
d257280201 | Human object interaction (HOI) detection plays a crucial role in human-centric scene understanding and serves as a fundamental building-block for many vision tasks. One generalizable and scalable strategy for HOI detection is to use weak supervision, learning from image-level annotations only. This is inherently challenging due to ambiguous human-object associations, large search space of detecting HOIs and highly noisy training signal. A promising strategy to address those challenges is to exploit knowledge from large-scale pretrained models (e.g., CLIP), but a direct knowledge distillation strategy (Liao et al., 2022) does not perform well on the weakly-supervised setting. In contrast, we develop a CLIP-guided HOI representation capable of incorporating the prior knowledge at both image level and HOI instance level, and adopt a self-taught mechanism to prune incorrect human-object associations. Experimental results on HICO-DET and V-COCO show that our method outperforms the previous works by a sizable margin, showing the efficacy of our HOI representation. Code is available at https://github.com/bobwan1995/Weakly-HOI. * Authors contributed equally and are listed in alphabetical order. | WEAKLY-SUPERVISED HOI DETECTION VIA PRIOR- GUIDED BI-LEVEL REPRESENTATION LEARNING |
d250627507 | We introduce a practical method to enforce partial differential equation (PDE) constraints for functions defined by neural networks (NNs), with a high degree of accuracy and up to a desired tolerance. We develop a differentiable PDEconstrained layer that can be incorporated into any NN architecture. Our method leverages differentiable optimization and the implicit function theorem to effectively enforce physical constraints. Inspired by dictionary learning, our model learns a family of functions, each of which defines a mapping from PDE parameters to PDE solutions. At inference time, the model finds an optimal linear combination of the functions in the learned family by solving a PDE-constrained optimization problem. Our method provides continuous solutions over the domain of interest that accurately satisfy desired physical constraints. Our results show that incorporating hard constraints directly into the NN architecture achieves much lower test error when compared to training on an unconstrained objective. | Published as a conference paper at ICLR 2023 LEARNING DIFFERENTIABLE SOLVERS FOR SYSTEMS WITH HARD CONSTRAINTS |
d231985467 | Neural Ordinary Differential Equations (NODEs), a framework of continuousdepth neural networks, have been widely applied, showing exceptional efficacy in coping with some representative datasets. Recently, an augmented framework has been successfully developed for conquering some limitations emergent in application of the original framework. Here we propose a new class of continuous-depth neural networks with delay, named as Neural Delay Differential Equations (ND-DEs), and, for computing the corresponding gradients, we use the adjoint sensitivity method to obtain the delayed dynamics of the adjoint. Since the differential equations with delays are usually seen as dynamical systems of infinite dimension possessing more fruitful dynamics, the NDDEs, compared to the NODEs, own a stronger capacity of nonlinear representations. Indeed, we analytically validate that the NDDEs are of universal approximators, and further articulate an extension of the NDDEs, where the initial function of the NDDEs is supposed to satisfy ODEs. More importantly, we use several illustrative examples to demonstrate the outstanding capacities of the NDDEs and the NDDEs with ODEs' initial value. Specifically, (1) we successfully model the delayed dynamics where the trajectories in the lower-dimensional phase space could be mutually intersected, while the traditional NODEs without any argumentation are not directly applicable for such modeling, and (2) we achieve lower loss and higher accuracy not only for the data produced synthetically by complex models but also for the real-world image datasets, i.e., CIFAR10, MNIST, and SVHN. Our results on the NDDEs reveal that appropriately articulating the elements of dynamical systems into the network design is truly beneficial to promoting the network performance. | Published as a conference paper at ICLR 2021 NEURAL DELAY DIFFERENTIAL EQUATIONS |
d235254386 | Explaining deep learning model inferences is a promising venue for scientific understanding, improving safety, uncovering hidden biases, evaluating fairness, and beyond, as argued by many scholars. One of the principal benefits of counterfactual explanations is allowing users to explore "what-if" scenarios through what does not and cannot exist in the data, a quality that many other forms of explanation such as heatmaps and influence functions are inherently incapable of doing. However, most previous work on generative explainability cannot disentangle important concepts effectively, produces unrealistic examples, or fails to retain relevant information. We propose a novel approach, DISSECT, that jointly trains a generator, a discriminator, and a concept disentangler to overcome such challenges using little supervision. DISSECT generates Concept Traversals (CTs), defined as a sequence of generated examples with increasing degrees of concepts that influence a classifier's decision. By training a generative model from a classifier's signal, DISSECT offers a way to discover a classifier's inherent "notion" of distinct concepts automatically rather than rely on user-predefined concepts. We show that DISSECT produces CTs that (1) disentangle several concepts, (2) are influential to a classifier's decision and are coupled to its reasoning due to joint training (3), are realistic, (4) preserve relevant information, and (5) are stable across similar inputs. We validate DISSECT on several challenging synthetic and realistic datasets where previous methods fall short of satisfying desirable criteria for interpretability and show that it performs consistently well. Finally, we present experiments showing applications of DISSECT for detecting potential biases of a classifier and identifying spurious artifacts that impact predictions. | DISSECT: DISENTANGLED SIMULTANEOUS EXPLA- NATIONS VIA CONCEPT TRAVERSALS |
d232404017 | Sequential deep learning models such as RNN, causal CNN and attention mechanism do not readily consume continuous-time information. Discretizing the temporal data, as we show, causes inconsistency even for simple continuous-time processes. Current approaches often handle time in a heuristic manner to be consistent with the existing deep learning architectures and implementations. In this paper, we provide a principled way to characterize continuous-time systems using deep learning tools. Notably, the proposed approach applies to all the major deep learning architectures and requires little modifications to the implementation. The critical insight is to represent the continuous-time system by composing neural networks with a temporal kernel, where we gain our intuition from the recent advancements in understanding deep learning with Gaussian process and neural tangent kernel. To represent the temporal kernel, we introduce the random feature approach and convert the kernel learning problem to spectral density estimation under reparameterization. We further prove the convergence and consistency results even when the temporal kernel is non-stationary, and the spectral density is misspecified. The simulations and real-data experiments demonstrate the empirical effectiveness of our temporal kernel approach in a broad range of settings. | A TEMPORAL KERNEL APPROACH FOR DEEP LEARN- ING WITH CONTINUOUS-TIME INFORMATION |
d1450294 | This paper addresses the visualisation of image classification models, learnt using deep Convolutional Networks (ConvNets). We consider two visualisation techniques, based on computing the gradient of the class score with respect to the input image. The first one generates an image, which maximises the class score [5], thus visualising the notion of the class, captured by a ConvNet. The second technique computes a class saliency map, specific to a given image and class. We show that such maps can be employed for weakly supervised object segmentation using classification ConvNets. Finally, we establish the connection between the gradient-based ConvNet visualisation methods and deconvolutional networks[13]. | Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps |
d257366012 | We propose a novel algorithm for offline reinforcement learning called Value Iteration with Perturbed Rewards (VIPeR), which amalgamates the pessimism principle with random perturbations of the value function. Most current offline RL algorithms explicitly construct statistical confidence regions to obtain pessimism via lower confidence bounds (LCB), which cannot easily scale to complex problems where a neural network is used to estimate the value functions. Instead, VIPeR implicitly obtains pessimism by simply perturbing the offline data multiple times with carefully-designed i.i.d. Gaussian noises to learn an ensemble of estimated state-action value functions and acting greedily with respect to the minimum of the ensemble. The estimated state-action values are obtained by fitting a parametric model (e.g., neural networks) to the perturbed datasets using gradient descent. As a result, VIPeR only needs O(1) time complexity for action selection, while LCB-based algorithms require at least Ω(K 2 ), where K is the total number of trajectories in the offline data. We also propose a novel data-splitting technique that helps remove a factor involving the log of the covering number in our bound. We prove that VIPeR yields a provable uncertainty quantifier with overparameterized neural networks and enjoys a bound on sub-optimality ofÕ(κH 5/2d / √ K), whered is the effective dimension, H is the horizon length and κ measures the distributional shift. We corroborate the statistical and computational efficiency of VIPeR with an empirical evaluation on a wide set of synthetic and real-world datasets. To the best of our knowledge, VIPeR is the first algorithm for offline RL that is provably efficient for general Markov decision processes (MDPs) with neural network function approximation. | VIPER: PROVABLY EFFICIENT ALGORITHM FOR OF- FLINE RL WITH NEURAL FUNCTION APPROXIMATION |
d238407943 | Computational methods for predicting the interface contacts between proteins come highly sought after for drug discovery as they can significantly advance the accuracy of alternative approaches, such as protein-protein docking, protein function analysis tools, and other computational methods for protein bioinformatics. In this work, we present the Geometric Transformer, a novel geometryevolving graph transformer for rotation and translation-invariant protein interface contact prediction, packaged within DeepInteract, an end-to-end prediction pipeline. DeepInteract predicts partner-specific protein interface contacts (i.e., inter-protein residue-residue contacts) given the 3D tertiary structures of two proteins as input. In rigorous benchmarks, DeepInteract, on challenging protein complex targets from the 13th and 14th CASP-CAPRI experiments as well as Docking Benchmark 5, achieves 14% and 1.1% top L/5 precision (L: length of a protein unit in a complex), respectively. In doing so, DeepInteract, with the Geometric Transformer as its graph-based backbone, outperforms existing methods for interface contact prediction in addition to other graph-based neural network backbones compatible with DeepInteract, thereby validating the effectiveness of the Geometric Transformer for learning rich relational-geometric features for downstream tasks on 3D protein structures. 1 | Published as a conference paper at ICLR 2022 GEOMETRIC TRANSFORMERS FOR PROTEIN INTERFACE CONTACT PREDICTION |
d232273772 | Although neural module networks have an architectural bias towards compositionality, they require gold standard layouts to generalize systematically in practice. When instead learning layouts and modules jointly, compositionality does not arise automatically and an explicit pressure is necessary for the emergence of layouts exhibiting the right structure. We propose to address this problem using iterated learning, a cognitive science theory of the emergence of compositional languages in nature that has primarily been applied to simple referential games in machine learning. Considering the layouts of module networks as samples from an emergent language, we use iterated learning to encourage the development of structure within this language. We show that the resulting layouts support systematic generalization in neural agents solving the more complex task of visual question-answering. Our regularized iterated learning method can outperform baselines without iterated learning on SHAPES-SyGeT (SHAPES Systematic Generalization Test), a new split of the SHAPES dataset we introduce to evaluate systematic generalization, and on CLOSURE, an extension of CLEVR also designed to test systematic generalization. We demonstrate superior performance in recovering ground-truth compositional program structure with limited supervision on both SHAPES-SyGeT and CLEVR. | Published as a conference paper at ICLR 2021 ITERATED LEARNING FOR EMERGENT SYSTEMATICITY IN VQA |
d257280374 | The use of well-disentangled representations offers many advantages for downstream tasks, e.g. an increased sample efficiency, or better interpretability. However, the quality of disentangled interpretations is often highly dependent on the choice of dataset-specific hyperparameters, in particular the regularization strength. To address this issue, we introduce DAVA, a novel training procedure for variational auto-encoders. DAVA completely alleviates the problem of hyperparameter selection. We compare DAVA to models with optimal hyperparameters. Without any hyperparameter tuning, DAVA is competitive on a diverse range of commonly used datasets. Underlying DAVA, we discover a necessary condition for unsupervised disentanglement, which we call PIPE. We demonstrate the ability of PIPE to positively predict the performance of downstream models in abstract reasoning. We also thoroughly investigate correlations with existing supervised and unsupervised metrics. The code is available at github.com/besterma/dava. | Published as a conference paper at ICLR 2023 DAVA: DISENTANGLING ADVERSARIAL VARIA- TIONAL AUTOENCODER |
d211126805 | We derive an unbiased estimator for expectations over discrete random variables based on sampling without replacement, which reduces variance as it avoids duplicate samples. We show that our estimator can be derived as the Rao-Blackwellization of three different estimators. Combining our estimator with RE-INFORCE, we obtain a policy gradient estimator and we reduce its variance using a built-in control variate which is obtained without additional model evaluations. The resulting estimator is closely related to other gradient estimators. Experiments with a toy problem, a categorical Variational Auto-Encoder and a structured prediction problem show that our estimator is the only estimator that is consistently among the best estimators in both high and low entropy settings. | Published as a conference paper at ICLR 2020 ESTIMATING GRADIENTS FOR DISCRETE RANDOM VARIABLES BY SAMPLING WITHOUT REPLACEMENT |
d221507913 | Data Poisoning attacks modify training data to maliciously control a model trained on such data. In this work, we focus on targeted poisoning attacks which cause a reclassification of an unmodified test image and as such breach model integrity. We consider a particularly malicious poisoning attack that is both "from scratch" and "clean label", meaning we analyze an attack that successfully works against new, randomly initialized models, and is nearly imperceptible to humans, all while perturbing only a small fraction of the training data. Previous poisoning attacks against deep neural networks in this setting have been limited in scope and success, working only in simplified settings or being prohibitively expensive for large datasets. The central mechanism of the new attack is matching the gradient direction of malicious examples. We analyze why this works, supplement with practical considerations. and show its threat to real-world practitioners, finding that it is the first poisoning method to cause targeted misclassification in modern deep networks trained from scratch on a full-sized, poisoned ImageNet dataset. Finally we demonstrate the limitations of existing defensive strategies against such an attack, concluding that data poisoning is a credible threat, even for large-scale deep learning systems. * Authors contributed equally. † Authors contributed equally. | WITCHES' BREW: INDUSTRIAL SCALE DATA POISON- ING VIA GRADIENT MATCHING |
d210023401 | In the Information Bottleneck (IB), when tuning the relative strength between compression and prediction terms, how do the two terms behave, and what's their relationship with the dataset and the learned representation? In this paper, we set out to answer these questions by studying multiple phase transitions in the IB objective: IB β [p(z|x)] = I(X; Z)−βI(Y ; Z) defined on the encoding distribution p(z|x) for input X, target Y and representation Z, where sudden jumps of dI(Y ;Z) dβ and prediction accuracy are observed with increasing β. We introduce a definition for IB phase transitions as a qualitative change of the IB loss landscape, and show that the transitions correspond to the onset of learning new classes. Using secondorder calculus of variations, we derive a formula that provides a practical condition for IB phase transitions, and draw its connection with the Fisher information matrix for parameterized models. We provide two perspectives to understand the formula, revealing that each IB phase transition is finding a component of maximum (nonlinear) correlation between X and Y orthogonal to the learned representation, in close analogy with canonical-correlation analysis (CCA) in linear settings. Based on the theory, we present an algorithm for discovering phase transition points. Finally, we verify that our theory and algorithm accurately predict phase transitions in categorical datasets, predict the onset of learning new classes and class difficulty in MNIST, and predict prominent phase transitions in CIFAR10. | PHASE TRANSITIONS FOR THE INFORMATION BOTTLE- NECK IN REPRESENTATION LEARNING |
d8739352 | Recently, we proposed to transform the outputs of each hidden neuron in a multilayer perceptron network to have zero output and zero slope on average, and use separate shortcut connections to model the linear dependencies instead. We continue the work by firstly introducing a third transformation to normalize the scale of the outputs of each hidden neuron, and secondly by analyzing the connections to second order optimization methods. We show that the transformations make a simple stochastic gradient behave closer to second-order optimization methods and thus speed up learning. This is shown both in theory and with experiments. The experiments on the third transformation show that while it further increases the speed of learning, it can also hurt performance by converging to a worse local optimum, where both the inputs and outputs of many hidden neurons are close to zero. | Pushing Stochastic Gradient towards Second-Order Methods -Backpropagation Learning with Transformations in Nonlinearities |
d3000562 | Neural language models predict the next token using a latent representation of the immediate token history. Recently, various methods for augmenting neural language models with an attention mechanism over a differentiable memory have been proposed. For predicting the next token, these models query information from a memory of the recent history which can facilitate learning mid-and long-range dependencies. However, conventional attention mechanisms used in memoryaugmented neural language models produce a single output vector per time step. This vector is used both for predicting the next token as well as for the key and value of a differentiable memory of a token history. In this paper, we propose a neural language model with a key-value attention mechanism that outputs separate representations for the key and value of a differentiable memory, as well as for encoding the next-word distribution. This model outperforms existing memoryaugmented neural language models on two corpora. Yet, we found that our method mainly utilizes a memory of the five most recent output representations. This led to the unexpected main finding that a much simpler model based only on the concatenation of recent output representations from previous time steps is on par with more sophisticated memory-augmented neural language models. | Published as a conference paper at ICLR 2017 FRUSTRATINGLY SHORT ATTENTION SPANS IN NEURAL LANGUAGE MODELING |
d8221720 | Question answering tasks have shown remarkable progress with distributed vector representation. In this paper, we investigate the recently proposed Facebook bAbI tasks which consist of twenty different categories of questions that require complex reasoning. Because the previous work on bAbI are all end-to-end models, errors could come from either an imperfect understanding of semantics or in certain steps of the reasoning. For clearer analysis, we propose two vector space models inspired by Tensor Product Representation (TPR) to perform knowledge encoding and logical reasoning based on common-sense inference. They together achieve near-perfect accuracy on all categories including positional reasoning and path finding that have proved difficult for most of the previous approaches. We hypothesize that the difficulties in these categories are due to the multi-relations in contrast to uni-relational characteristic of other categories. Our exploration sheds light on designing more sophisticated dataset and moving one step toward integrating transparent and interpretable formalism of TPR into existing learning paradigms. * This research was conducted while the first author held a summer internship in Microsoft Research, Redmond, and the last author was a Visiting Researcher there. | Published as a conference paper at ICLR 2016 REASONING IN VECTOR SPACE: AN EXPLORATORY STUDY OF QUESTION ANSWERING |
d238531802 | Despite being widely used, face recognition models suffer from bias: the probability of a false positive (incorrect face match) strongly depends on sensitive attributes such as the ethnicity of the face. As a result, these models can disproportionately and negatively impact minority groups, particularly when used by law enforcement. The majority of bias reduction methods have several drawbacks: they use an end-toend retraining approach, may not be feasible due to privacy issues, and often reduce accuracy. An alternative approach is post-processing methods that build fairer decision classifiers using the features of pre-trained models, thus avoiding the cost of retraining. However, they still have drawbacks: they reduce accuracy (AGENDA, PASS, FTC), or require retuning for different false positive rates (FSN). In this work, we introduce the Fairness Calibration (FairCal) method, a post-training approach that simultaneously: (i) increases model accuracy (improving the stateof-the-art), (ii) produces fairly-calibrated probabilities, (iii) significantly reduces the gap in the false positive rates, (iv) does not require knowledge of the sensitive attribute, and (v) does not require retraining, training an additional model, or retuning. We apply it to the task of Face Verification, and obtain state-of-the-art results with all the above advantages.Published as a conference paper at ICLR 2022 1. Accuracy: FairCal achieves state-of-the-art accuracy(Table 2)for face verification on two large datasets.2. Fairness-calibration: Our face verification classifier outputs state-of-the-art fairnesscalibrated probabilities without knowledge of the sensitive attribute(Table 3).3. Predictive equality: Our method reduces the gap in FPRs across sensitive attributes (Table 4). For example, inFigure 2, at a Global FPR of 5% using the baseline method Black people are 15X more likely to false match than white people. Our method reduces this to 1.2X (while SOTA for post-hoc methods is 1.7X).4.No sensitive attribute required: Our approach does not require the sensitive attribute, neither at training nor at test time. In fact, it outperforms models that use this knowledge. | FAIRCAL: FAIRNESS CALIBRATION FOR FACE VERIFI- CATION |
d12072571 | Identifying acoustic events from a continuously streaming audio source is of interest for many applications including environmental monitoring for basic research. In this scenario neither different event classes are known nor what distinguishes one class from another. Therefore, an unsupervised feature learning method for exploration of audio data is presented in this paper. It incorporates the two following novel contributions: First, an audio frame predictor based on a Convolutional LSTM autoencoder is demonstrated, which is used for unsupervised feature extraction. Second, a training method for autoencoders is presented, which leads to distinct features by amplifying event similarities. In comparison to standard approaches, the features extracted from the audio frame predictor trained with the novel approach show 13 % better results when used with a classifier and 36 % better results when used for clustering. | Workshop track -ICLR 2017 UNSUPERVISED FEATURE LEARNING FOR AUDIO ANALYSIS |
d222178254 | Although deep learning has achieved appealing results on several machine learning tasks, most of the models are deterministic at inference, limiting their application to single-modal settings. We propose a novel general-purpose framework for conditional generation in multimodal spaces, that uses latent variables to model generalizable learning patterns while minimizing a family of regression cost functions. At inference, the latent variables are optimized to find optimal solutions corresponding to multiple output modes. Compared to existing generative solutions, our approach demonstrates faster and stable convergence, and can learn better representations for downstream tasks. Importantly, it provides a simple generic model that can beat highly engineered pipelines tailored using domain expertise on a variety of tasks, while generating diverse outputs. Our codes will be released. | CONDITIONAL GENERATIVE MODELING VIA LEARN- ING THE LATENT SPACE |
d207870430 | We introduce kNN-LMs, which extend a pre-trained neural language model (LM) by linearly interpolating it with a k-nearest neighbors (kNN) model. The nearest neighbors are computed according to distance in the pre-trained LM embedding space, and can be drawn from any text collection, including the original LM training data. Applying this augmentation to a strong WIKITEXT-103 LM, with neighbors drawn from the original training set, our kNN-LM achieves a new stateof-the-art perplexity of 15.79 -a 2.9 point improvement with no additional training. We also show that this approach has implications for efficiently scaling up to larger training sets and allows for effective domain adaptation, by simply varying the nearest neighbor datastore, again without further training. Qualitatively, the model is particularly helpful in predicting rare patterns, such as factual knowledge. Together, these results strongly suggest that learning similarity between sequences of text is easier than predicting the next word, and that nearest neighbor search is an effective approach for language modeling in the long tail. . Pointer sentinel mixture models. ICLR, 2017. Tomáš Mikolov, Martin Karafiát, Lukáš Burget, JanČernockỳ, and Sanjeev Khudanpur. Recurrent neural network based language model. In Eleventh annual conference of the international speech communication association, 2010. A. Emin Orhan. A simple cache model for image recognition. In NeurIPS, 2018. Nicolas Papernot and Patrick McDaniel. Deep k-nearest neighbors: Towards confident, interpretable and robust deep learning. arXiv preprint arXiv:1803.04765, 2018.Ofir Press and Lior Wolf. Using the output embedding to improve language models. In ICLR, 2017. machine translation of rare words with subword units. arXiv preprint arXiv:1508.07909, 2015. Catanzaro. Megatron-lm: Training multi-billion parameter language models using gpu model parallelism. arXiv preprint arXiv:1909.08053, 2019. | Published as a conference paper at ICLR 2020 GENERALIZATION THROUGH MEMORIZATION: NEAREST NEIGHBOR LANGUAGE MODELS |
d202889315 | Group convolutional neural networks (G-CNNs) can be used to improve classical CNNs by equipping them with the geometric structure of groups. Central in the success of G-CNNs is the lifting of feature maps to higher dimensional disentangled representations, in which data characteristics are effectively learned, geometric data-augmentations are made obsolete, and predictable behavior under geometric transformations (equivariance) is guaranteed via group theory. Currently, however, the practical implementations of G-CNNs are limited to either discrete groups (that leave the grid intact) or continuous compact groups such as rotations (that enable the use of Fourier theory). In this paper we lift these limitations and propose a modular framework for the design and implementation of G-CNNs for arbitrary Lie groups. In our approach the differential structure of Lie groups is used to expand convolution kernels in a generic basis of B-splines that is defined on the Lie algebra. This leads to a flexible framework that enables localized, atrous, and deformable convolutions in G-CNNs by means of respectively localized, sparse and non-uniform B-spline expansions. The impact and potential of our approach is studied on two benchmark datasets: cancer detection in histopathology slides in which rotation equivariance plays a key role and facial landmark localization in which scale equivariance is important. In both cases, G-CNN architectures outperform their classical 2D counterparts and the added value of atrous and localized group convolutions is studied in detail. arXiv:1909.12057v1 [cs.LG] 26 Sep 2019 ArXiv preprint. Under review. (a) Pattern of local orientations (b) Density on SE(2) Figure 1: (a) A face descriptor in terms of low level features (e.g. edges) in a pattern of local orientations (elements in SE(2)) relative to an origin e and (b) the same pattern embed as a density on SE(2) that represents idealized neuronal activations in a G-CNN feature map. Representing low-level features via features maps on groups, as is done in G-CNNs, is also motivated by the findings of Hubel & Wiesel (1959) and Bosking et al. (1997) on the organization of orientation sensitive simple cells in the primary visual cortex V1. These findings are mathematically modeled by sub-Riemannian geometry on Lie groups (Petitot, ). In recent work Montobbio et al. (2019) show that such advanced V1 modeling geometries emerge in specific CNN architectures and in Ecker et al. (2019) the relation between group structure and the organization of V1 is explicitly employed to effectively recover actual V1 neuronal activities from stimuli by means of G-CNNs. G-CNNs are well motivated from both a mathematical point of view (Cohen et al., 2018a; Kondor & Trivedi, 2018) and neuro-psychological/neuro-mathematical point of view and their improvement over classical CNNs is convincingly demonstrated by the growing body of G-CNN literature (see Sec.2). However, their practical implementations are limited to either discrete groups (that leave the grid intact) or continuous, (locally) compact, unimodular groups such as roto-translations (that enable the use of Fourier theory). In this paper we lift these limitations and propose a framework for the design and implementation of G-CNNs for arbitrary Lie groups.The proposed approach for G-CNNs relies on a definition of B-splines on Lie groups which we use to expand and sample group convolution kernels. B-splines are piece-wise polynomials with local support and are classically defined on flat Euclidean spaces R d . In this paper we generalize B-splines to Lie groups and formulate a definition using the differential structure of Lie groups in which Bsplines are essentially defined on the (flat) vector space of the Lie algebra obtained by the logarithmic map. The result is a flexible framework for B-splines on arbitrary Lie groups and it enables the construction of G-CNNs with properties that cannot be achieved via traditional Fourier-type basis expansion methods. Such properties include localized, atrous, and deformable convolutions in G-CNNs by means of respectively localized, sparse and non-uniform B-spline expansions.Although concepts described in this paper apply to arbitrary Lie groups, we here concentrate on the analysis of data that lives on R d and consider G-CNNs for affine groups G = R d H that are the semi-direct product of the translation group with a Lie group H that acts on R d . As such, only a few core definitions about the Lie group H (group product, inverse, Log, and action on R d ) need to be implemented in order to build full G-CNNs that are locally equivariant to the transformations in H.The impact and potential of our approach is studied on two datasets in which respectively rotation and scale equivariance plays a key role: cancer detection in histopathology slides (PCam dataset) and facial landmark localization (CelebA dataset). In both cases G-CNNs out-perform their classical 2D counterparts and the added value of atrous and localized G-convolutions is studied in detail. | ArXiv preprint. Under review. B-SPLINE CNNS ON LIE GROUPS |
d252683782 | Unsupervised reinforcement learning (URL) poses a promising paradigm to learn useful behaviors in a task-agnostic environment without the guidance of extrinsic rewards to facilitate the fast adaptation of various downstream tasks. Previous works focused on the pre-training in a model-free manner while lacking the study of transition dynamics modeling that leaves a large space for the improvement of sample efficiency in downstream tasks. To this end, we propose an Efficient Unsupervised reinforCement Learning framework with multi-choIce Dynamics model (EUCLID), which introduces a novel model-fused paradigm to jointly pretrain the dynamics model and unsupervised exploration policy in the pre-training phase, thus better leveraging the environmental samples and improving the downstream task sampling efficiency. However, constructing a generalizable model which captures the local dynamics under different behaviors remains a challenging problem. We introduce the multi-choice dynamics model that covers different local dynamics under different behaviors concurrently, which uses different heads to learn the state transition under different behaviors during unsupervised pretraining and selects the most appropriate head for prediction in the downstream task. Experimental results in the manipulation and locomotion domains demonstrate that EUCLID achieves state-of-the-art performance with high sample efficiency, basically solving the state-based URLB benchmark and reaching a mean normalized score of 104.0±1.2% in downstream tasks with 100k fine-tuning steps, which is equivalent to DDPG's performance at 2M interactive steps with 20× more data. More visualization videos are released on our homepage. | EUCLID: TOWARDS EFFICIENT UNSUPERVISED RE- INFORCEMENT LEARNING WITH MULTI-CHOICE DY- NAMICS MODEL |
d3632319 | We present NeuroSAT, a message passing neural network that learns to solve SAT problems after only being trained as a classifier to predict satisfiability. Although it is not competitive with state-of-the-art SAT solvers, NeuroSAT can solve problems that are substantially larger and more difficult than it ever saw during training by simply running for more iterations. Moreover, NeuroSAT generalizes to novel distributions; after training only on random SAT problems, at test time it can solve SAT problems encoding graph coloring, clique detection, dominating set, and vertex cover problems, all on a range of distributions over small random graphs. | LEARNING A SAT SOLVER FROM SINGLE-BIT SUPER- VISION |
d247763065 | Many existing conditional score-based data generation methods utilize Bayes' theorem to decompose the gradients of a log posterior density into a mixture of scores. These methods facilitate the training procedure of conditional score models, as a mixture of scores can be separately estimated using a score model and a classifier. However, our analysis indicates that the training objectives for the classifier in these methods may lead to a serious score mismatch issue, which corresponds to the situation that the estimated scores deviate from the true ones. Such an issue causes the samples to be misled by the deviated scores during the diffusion process, resulting in a degraded sampling quality. To resolve it, we formulate a novel training objective, called Denoising Likelihood Score Matching (DLSM) loss, for the classifier to match the gradients of the true log likelihood density. Our experimental evidence shows that the proposed method outperforms the previous methods on both Cifar-10 and Cifar-100 benchmarks noticeably in terms of several key evaluation metrics. We thus conclude that, by adopting DLSM, the conditional scores can be accurately modeled, and the effect of the score mismatch issue is alleviated. * Work done during an internship at Mediatek Inc. | Published as a conference paper at ICLR 2022 DENOISING LIKELIHOOD SCORE MATCHING FOR CONDITIONAL SCORE-BASED DATA GENERATION |
d208915826 | The paper presents an O(N log N )-implementation of t-SNE -an embedding technique that is commonly used for the visualization of high-dimensional data in scatter plots and that normally runs in O(N 2 ). The new implementation uses vantage-point trees to compute sparse pairwise similarities between the input data objects, and it uses a variant of the Barnes-Hut algorithm to approximate the forces between the corresponding points in the embedding. Our experiments show that the new algorithm, called Barnes-Hut-SNE, leads to substantial computational advantages over standard t-SNE, and that it makes it possible to learn embeddings of data sets with millions of objects. | Barnes-Hut-SNE |
d248300043 | Adversarial Propagation (AdvProp) is an effective way to improve recognition models, leveraging adversarial examples. Nonetheless, AdvProp suffers from the extremely slow training speed, mainly because: a) extra forward and backward passes are required for generating adversarial examples; b) both original samples and their adversarial counterparts are used for training (i.e., 2× data). In this paper, we introduce Fast AdvProp, which aggressively revamps AdvProp's costly training components, rendering the method nearly as cheap as the vanilla training. Specifically, our modifications in Fast AdvProp are guided by the hypothesis that disentangled learning with adversarial examples is the key for performance improvements, while other training recipes (e.g., paired clean and adversarial training samples, multi-step adversarial attackers) could be largely simplified. Our empirical results show that, compared to the vanilla training baseline, Fast AdvProp is able to further model performance on a spectrum of visual benchmarks, without incurring extra training cost. Additionally, our ablations find Fast AdvProp scales better if larger models are used, is compatible with existing data augmentation methods (i.e., Mixup and CutMix), and can be easily adapted to other recognition tasks like object detection. The code is available here: | Published as a conference paper at ICLR 2022 FAST ADVPROP |
d254823436 | The success of deep learning heavily relies on large-scale data with comprehensive labels, which is more expensive and time-consuming to fetch in 3D compared to 2D images or natural languages. This promotes the potential of utilizing models pretrained with data more than 3D as teachers for cross-modal knowledge transferring. In this paper, we revisit masked modeling in a unified fashion of knowledge distillation, and we show that foundational Transformers pretrained with 2D images or natural languages can help self-supervised 3D representation learning through training Autoencoders as Cross-Modal Teachers (ACT). The pretrained Transformers are transferred as cross-modal 3D teachers using discrete variational autoencoding self-supervision, during which the Transformers are frozen with prompt tuning for better knowledge inheritance. The latent features encoded by the 3D teachers are used as the target of masked point modeling, wherein the dark knowledge is distilled to the 3D Transformer students as foundational geometry understanding. Our ACT pretrained 3D learner achieves state-of-the-art generalization capacity across various downstream benchmarks, e.g., 88.21% overall accuracy on ScanObjectNN. Codes have been released at https://github.com/RunpeiDong/ACT. † Corresponding authors. ‡ Work partially done during the internship of Runpei Dong | Published as a conference paper at ICLR 2023 AUTOENCODERS AS CROSS-MODAL TEACHERS: CAN PRETRAINED 2D IMAGE TRANSFORMERS HELP 3D REPRESENTATION LEARNING? |
d35615696 | Whereas it is believed that techniques such as Adam, batch normalization and, more recently, SeLU nonlinearities "solve" the exploding gradient problem, we show that this is not the case in general and that in a range of popular MLP architectures, exploding gradients exist and that they limit the depth to which networks can be effectively trained, both in theory and in practice. We explain why exploding gradients occur and highlight the collapsing domain problem, which can arise in architectures that avoid exploding gradients. ResNets have significantly lower gradients and thus can circumvent the exploding gradient problem, enabling the effective training of much deeper networks, which we show is a consequence of a surprising mathematical property. By noticing that any neural network is a residual network, we devise the residual trick, which reveals that introducing skip connections simplifies the network mathematically, and that this simplicity may be the major cause for their success. | GRADIENTS EXPLODE -DEEP NETWORKS ARE SHALLOW -RESNET EXPLAINED |
d210699231 | Training with larger number of parameters while keeping fast iterations is an increasingly adopted strategy and trend for developing better performing Deep Neural Network (DNN) models. This necessitates increased memory footprint and computational requirements for training. Here we introduce a novel methodology for training deep neural networks using 8-bit floating point (FP8) numbers. Reduced bit precision allows for a larger effective memory and increased computational speed. We name this method Shifted and Squeezed FP8 (S2FP8). We show that, unlike previous 8-bit precision training methods, the proposed method works out-of-the-box for representative models: ResNet-50, Transformer and NCF. The method can maintain model accuracy without requiring fine-tuning loss scaling parameters or keeping certain layers in single precision. We introduce two learnable statistics of the DNN tensors -shifted and squeezed factors that are used to optimally adjust the range of the tensors in 8-bits, thus minimizing the loss in information due to quantization. * Work performed during an internship at Intel † Equal contribution arXiv:2001.05674v1 [cs.LG] | Published as a conference paper at ICLR 2020 SHIFTED AND SQUEEZED 8-BIT FLOATING POINT FOR- MAT FOR LOW-PRECISION TRAINING OF DEEP NEU- RAL NETWORKS |
d257050386 | Spiking Neural Networks (SNNs) have attracted great attention due to their distinctive characteristics of low power consumption and temporal information processing. ANN-SNN conversion, as the most commonly used training method for applying SNNs, can ensure that converted SNNs achieve comparable performance to ANNs on large-scale datasets. However, the performance degrades severely under low quantities of time-steps, which hampers the practical applications of SNNs to neuromorphic chips. In this paper, instead of evaluating different conversion errors and then eliminating these errors, we define an offset spike to measure the degree of deviation between actual and desired SNN firing rates. We perform a detailed analysis of offset spike and note that the firing of one additional (or one less) spike is the main cause of conversion errors. Based on this, we propose an optimization strategy based on shifting the initial membrane potential and we theoretically prove the corresponding optimal shifting distance for calibrating the spike. In addition, we also note that our method has a unique iterative property that enables further reduction of conversion errors. The experimental results show that our proposed method achieves state-of-the-art performance on CIFAR-10, CIFAR-100, and ImageNet datasets. For example, we reach a top-1 accuracy of 67.12% on ImageNet when using 6 time-steps. To the best of our knowledge, this is the first time an ANN-SNN conversion has been shown to simultaneously achieve high accuracy and ultralow latency on complex datasets. | Published as a conference paper at ICLR 2023 BRIDGING THE GAP BETWEEN ANNS AND SNNS BY CALIBRATING OFFSET SPIKES |
d252668904 | Crossmodal knowledge distillation (KD) extends traditional knowledge distillation to the area of multimodal learning and demonstrates great success in various applications. To achieve knowledge transfer across modalities, a pretrained network from one modality is adopted as the teacher to provide supervision signals to a student network learning from another modality. In contrast to the empirical success reported in prior works, the working mechanism of crossmodal KD remains a mystery. In this paper, we present a thorough understanding of crossmodal KD. We begin with two case studies and demonstrate that KD is not a universal cure in crossmodal knowledge transfer. We then present the modality Venn diagram (MVD) to understand modality relationships and the modality focusing hypothesis (MFH) revealing the decisive factor in the efficacy of crossmodal KD. Experimental results on 6 multimodal datasets help justify our hypothesis, diagnose failure cases, and point directions to improve crossmodal knowledge transfer in the future. 1 ;Tang et al., 2020;are in search for the answer in the context of model capacity difference and architecture difference. However, the analysis for the third scenario, KD under modality difference or formally crossmodal KD, remains an open problem. This work aims to fill this gap and for the first time provides a comprehensive analysis of crossmodal KD. Our major contributions are the following:• We evaluate crossmodal KD on a few multimodal tasks and find surprisingly that teacher performance does not always positively correlate with student performance. • To explore the cause of performance mismatch in crossmodal KD, we adopt the modality Venn diagram (MVD) to understand modality relationships and formally define modalitygeneral decisive features and modality-specific decisive features. • We present the modality focusing hypothesis (MFH) that provides an explanation of when crossmodal KD is effective. We hypothesize that modality-general decisive features are the crucial factor that determines the efficacy of crossmodal KD. • We conduct experiments on 6 multimodal datasets (i.e., synthetic Gaussian, AV-MNIST, RAVDESS, VGGSound, NYU Depth V2, and MM-IMDB). The results validate the proposed MFH and provide insights on how to improve crossmodal KD.Geoffrey Hinton, Oriol Vinyals, Jeff Dean, et al. Distilling the knowledge in a neural network. arXiv preprint arXiv:1503.02531, 2(7), 2015. | THE MODALITY FOCUSING HYPOTHESIS: TOWARDS UNDERSTANDING CROSSMODAL KNOWLEDGE DISTIL- LATION |
d248218711 | This paper studies node classification in the inductive setting, i.e., aiming to learn a model on labeled training graphs and generalize it to infer node labels on unlabeled test graphs. This problem has been extensively studied with graph neural networks (GNNs) by learning effective node representations, as well as traditional structured prediction methods for modeling the structured output of node labels, e.g., conditional random fields (CRFs). In this paper, we present a new approach called the Structured Proxy Network (SPN), which combines the advantages of both worlds. SPN defines flexible potential functions of CRFs with GNNs. However, learning such a model is nontrivial as it involves optimizing a maximin game with high-cost inference. Inspired by the underlying connection between joint and marginal distributions defined by Markov networks, we propose to solve an approximate version of the optimization problem as a proxy, which yields a nearoptimal solution, making learning more efficient. Extensive experiments on two settings show that our approach outperforms many competitive baselines 1 . * Equal contribution. | Published as a conference paper at ICLR 2022 NEURAL STRUCTURED PREDICTION FOR INDUCTIVE NODE CLASSIFICATION |
d207847681 | Graph representation learning for hypergraphs can be used to extract patterns among higher-order interactions that are critically important in many real world problems. Current approaches designed for hypergraphs, however, are unable to handle different types of hypergraphs and are typically not generic for various learning tasks. Indeed, models that can predict variable-sized heterogeneous hyperedges have not been available. Here we develop a new self-attention based graph neural network called Hyper-SAGNN applicable to homogeneous and heterogeneous hypergraphs with variable hyperedge sizes. We perform extensive evaluations on multiple datasets, including four benchmark network datasets and two single-cell Hi-C datasets in genomics. We demonstrate that Hyper-SAGNN significantly outperforms the state-of-the-art methods on traditional tasks while also achieving great performance on a new task called outsider identification. Hyper-SAGNN will be useful for graph representation learning to uncover complex higher-order interactions in different applications.Corresponding Author (node) Coauthor (node) Coauthorship (hyperedge)Figure 1: An example of the co-authorship hypergraph. Here authors are represented as nodes (in dark blue and light blue) and coauthorships are represented as hyperedges.In this work, we developed a new self-attention based graph neural network, called Hyper-SAGNN that can work with both homogeneous and heterogeneous hypergraphs with variable hyperedge size. Using the same datasets in the DHNE paper(Tu et al., 2018), we demonstrated the advantage of Hyper-SAGNN over DHNE in multiple tasks. We further tested the effectiveness of the method in predicting edges and hyperedges and showed that the model can achieve better performance from the multi-tasking setting. We also formulated a novel task called outsider identification and showed that Hyper-SAGNN performs strongly. Importantly, as an application of Hyper-SAGNN to single-cell genomics, we were able to learn the embeddings for the most recently produced single-cell Hi-C (scHi-C) datasets to uncover the clustering of cells based on their 3D genome structure(Ramani et al., 2017;Nagano et al., 2017). We showed that Hyper-SAGNN achieved improved results in identifying distinct cell populations as compared to existing scHi-C clustering methods. Taken together, Hyper-SAGNN can significantly outperform the state-of-the-art methods and can be applied to a wide range of hypergraphs for different applications. | Hyper-SAGNN: a self-attention based graph neural network for hypergraphs A PREPRINT |
d238407713 | Though recent works have developed methods that can generate estimates (or imputations) of the missing entries in a dataset to facilitate downstream analysis, most depend on assumptions that may not align with real-world applications and could suffer from poor performance in subsequent tasks such as classification. This is particularly true if the data have large missingness rates or a small sample size. More importantly, the imputation error could be propagated into the prediction step that follows, which may constrain the capabilities of the prediction model. In this work, we introduce the gradient importance learning (GIL) method to train multilayer perceptrons (MLPs) and long short-term memories (LSTMs) to directly perform inference from inputs containing missing values without imputation. Specifically, we employ reinforcement learning (RL) to adjust the gradients used to train these models via back-propagation. This allows the model to exploit the underlying information behind missingness patterns. We test the approach on real-world time-series (i.e., MIMIC-III), tabular data obtained from an eye clinic, and a standard dataset (i.e., MNIST), where our imputation-free predictions outperform the traditional two-step imputation-based predictions using state-of-the-art imputation methods. | Published as a conference paper at ICLR 2022 GRADIENT IMPORTANCE LEARNING FOR INCOMPLETE OBSERVATIONS |
d253553811 | We tackle the problem of Selective Classification where the objective is to achieve the best performance on a predetermined ratio (coverage) of the dataset. Recent state-of-the-art selective methods come with architectural changes either via introducing a separate selection head or an extra abstention logit. In this paper, we challenge the aforementioned methods. The results suggest that the superior performance of state-of-the-art methods is owed to training a more generalizable classifier rather than their proposed selection mechanisms. We argue that the best performing selection mechanism should instead be rooted in the classifier itself. Our proposed selection strategy uses the classification scores and achieves better results by a significant margin, consistently, across all coverages and all datasets, without any added compute cost. Furthermore, inspired by semi-supervised learning, we propose an entropy-based regularizer that improves the performance of selective classification methods. Our proposed selection mechanism with the proposed entropy-based regularizer achieves new state-of-the-art results. | Published as a conference paper at ICLR 2023 TOWARDS BETTER SELECTIVE CLASSIFICATION |
d247411040 | Vision Transformer (ViT) has recently demonstrated promise in computer vision problems. However, unlike Convolutional Neural Networks (CNN), it is known that the performance of ViT saturates quickly with depth increasing, due to the observed attention collapse or patch uniformity. Despite a couple of empirical solutions, a rigorous framework studying on this scalability issue remains elusive. In this paper, we first establish a rigorous theory framework to analyze ViT features from the Fourier spectrum domain. We show that the self-attention mechanism inherently amounts to a low-pass filter, which indicates when ViT scales up its depth, excessive low-pass filtering will cause feature maps to only preserve their Direct-Current (DC) component. We then propose two straightforward yet effective techniques to mitigate the undesirable low-pass limitation. The first technique, termed AttnScale, decomposes a self-attention block into low-pass and high-pass components, then rescales and combines these two filters to produce an all-pass self-attention matrix. The second technique, termed FeatScale, re-weights feature maps on separate frequency bands to amplify the high-frequency signals. Both techniques are efficient and hyperparameter-free, while effectively overcoming relevant ViT training artifacts such as attention collapse and patch uniformity. By seamlessly plugging in our techniques to multiple ViT variants, we demonstrate that they consistently help ViTs benefit from deeper architectures, bringing up to 1.1% performance gains "for free" (e.g., with little parameter overhead). We publicly release our codes and pre-trained models at | ANTI-OVERSMOOTHING IN DEEP VISION TRANS- FORMERS VIA THE FOURIER DOMAIN ANALYSIS: FROM THEORY TO PRACTICE |
d3603886 | Researches on deep neural networks with discrete parameters and their deployment in embedded systems have been active and promising topics. Although previous works have successfully reduced precision in inference, transferring both training and inference processes to low-bitwidth integers has not been demonstrated simultaneously. In this work, we develop a new method termed as "WAGE" to discretize both training and inference, where weights (W), activations (A), gradients (G) and errors (E) among layers are shifted and linearly constrained to low-bitwidth integers. To perform pure discrete dataflow for fixed-point devices, we further replace batch normalization by a constant scaling layer and simplify other components that are arduous for integer implementation. Improved accuracies can be obtained on multiple datasets, which indicates that WAGE somehow acts as a type of regularization. Empirically, we demonstrate the potential to deploy training in hardware systems such as integer-based deep learning accelerators and neuromorphic chips with comparable accuracy and higher energy efficiency, which is crucial to future AI applications in variable scenarios with transfer and continual learning demands.Published as a conference paper at ICLR 2018 to achieve ternary weights, 8-bit integers for activations and gradients accumulation. Secondly, for operations, batch normalization(Ioffe & Szegedy, 2015)is replaced by a constant scaling factor. Other techniques for fine-tuning such as SGD optimizer with momentum and L2 regularization are simplified or abandoned with little performance degradation. Considering the overall bidirectional propagation, we completely streamline inference into accumulate-compare cycles and training into low-bitwidth multiply-accumulate (MAC) cycles with alignment operations, respectively.We heuristically explore the bitwidth requirements of integers for error computation and gradient accumulation, which have rarely been discussed in previous works. Experiments indicate that it is the relative values (orientations) rather than absolute values (orders of magnitude) in error that guides previous layers to converge. Moreover, small values have negligible effects on previous orientations though propagated layer by layer, which can be partially discarded in quantization. We leverage these phenomena and use an orientation-preserved shifting operation to constrain errors. As for the gradient accumulation, though weights are quantized to ternary values in inference, a relatively higher bitwidth is indispensable to store and accumulate gradient updates.The proposed framework is evaluated on MNIST, CIFAR10, SVHN, ImageNet datasets. Comparing to those who only discretize weights and activations at inference time, it has comparable accuracy and can further alleviate overfitting, indicating some type of regularization. WAGE produces pure bidirectional low-precision integer dataflow for DNNs, which can be applied for training and inference in dedicated hardware neatly. We publish the code on GitHub 1 . | Published as a conference paper at ICLR 2018 TRAINING AND INFERENCE WITH INTEGERS IN DEEP NEURAL NETWORKS |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.