ICLR 2021 论文列表

In recent years, great success has been witnessed in building problem-specific deep networks from unrolling iterative algorithms, for solving inverse problems and beyond. Unrolling is believed to incorporate the model-based prior with the learning capacity of deep learning. This paper revisits \textit{the role of unrolling as a design approach for deep networks}: to what extent its resulting special architecture is superior, and can we find better? Using LISTA for sparse recovery as a representative example, we conduct the first thorough \textit{design space study} for the unrolled models. Among all possible variations, we focus on extensively varying the connectivity patterns and neuron types, leading to a gigantic design space arising from LISTA. To efficiently explore this space and identify top performers, we leverage the emerging tool of neural architecture search (NAS). We carefully examine the searched top architectures in a number of settings, and are able to discover networks that consistently better than LISTA. We further present more visualization and analysis to ``open the black box", and find that the searched top architectures demonstrate highly consistent and potentially transferable patterns. We hope our study to spark more reflections and explorations on how to better mingle model-based optimization prior and data-driven learning.

Generative adversarial networks (GANs) have emerged as a powerful unsupervised method to model the statistical patterns of real-world data sets, such as natural images. These networks are trained to map random inputs in their latent space to new samples representative of the learned data. However, the structure of the latent space is hard to intuit due to its high dimensionality and the non-linearity of the generator, which limits the usefulness of the models. Understanding the latent space requires a way to identify input codes for existing real-world images (inversion), and a way to identify directions with known image transformations (interpretability). Here, we use a geometric framework to address both issues simultaneously. We develop an architecture-agnostic method to compute the Riemannian metric of the image manifold created by GANs. The eigen-decomposition of the metric isolates axes that account for different levels of image variability. An empirical analysis of several pretrained GANs shows that image variation around each position is concentrated along surprisingly few major axes (the space is highly anisotropic) and the directions that create this large variation are similar at different positions in the space (the space is homogeneous). We show that many of the top eigenvectors correspond to interpretable transforms in the image space, with a substantial part of eigenspace corresponding to minor transforms which could be compressed out. This geometric understanding unifies key previous results related to GAN interpretability. We show that the use of this metric allows for more efficient optimization in the latent space (e.g. GAN inversion) and facilitates unsupervised discovery of interpretable axes. Our results illustrate that defining the geometry of the GAN image manifold can serve as a general framework for understanding GANs.

Federated learning is typically approached as an optimization problem, where the goal is to minimize a global loss function by distributing computation across client devices that possess local data and specify different parts of the global objective. We present an alternative perspective and formulate federated learning as a posterior inference problem, where the goal is to infer a global posterior distribution by having client devices each infer the posterior of their local data. While exact inference is often intractable, this perspective provides a principled way to search for global optima in federated settings. Further, starting with the analysis of federated quadratic objectives, we develop a computation- and communication-efficient approximate posterior inference algorithm—federated posterior averaging (FedPA). Our algorithm uses MCMC for approximate inference of local posteriors on the clients and efficiently communicates their statistics to the server, where the latter uses them to refine a global estimate of the posterior mode. Finally, we show that FedPA generalizes federated averaging (FedAvg), can similarly benefit from adaptive optimizers, and yields state-of-the-art results on four realistic and challenging benchmarks, converging faster, to better optima.

Continual (sequential) training and multitask (simultaneous) training are often attempting to solve the same overall objective: to find a solution that performs well on all considered tasks. The main difference is in the training regimes, where continual learning can only have access to one task at a time, which for neural networks typically leads to catastrophic forgetting. That is, the solution found for a subsequent task does not perform well on the previous ones anymore. However, the relationship between the different minima that the two training regimes arrive at is not well understood. What sets them apart? Is there a local structure that could explain the difference in performance achieved by the two different schemes? Motivated by recent work showing that different minima of the same task are typically connected by very simple curves of low error, we investigate whether multitask and continual solutions are similarly connected. We empirically find that indeed such connectivity can be reliably achieved and, more interestingly, it can be done by a linear path, conditioned on having the same initialization for both. We thoroughly analyze this observation and discuss its significance for the continual learning process. Furthermore, we exploit this finding to propose an effective algorithm that constrains the sequentially learned minima to behave as the multitask solution. We show that our method outperforms several state of the art continual learning algorithms on various vision benchmarks.

In this work we consider data-driven optimization problems where one must maximize a function given only queries at a fixed set of points. This problem setting emerges in many domains where function evaluation is a complex and expensive process, such as in the design of materials, vehicles, or neural network architectures. Because the available data typically only covers a small manifold of the possible space of inputs, a principal challenge is to be able to construct algorithms that can reason about uncertainty and out-of-distribution values, since a naive optimizer can easily exploit an estimated model to return adversarial inputs. We propose to tackle the MBO problem by leveraging the normalized maximum-likelihood (NML) estimator, which provides a principled approach to handling uncertainty and out-of-distribution inputs. While in the standard formulation NML is intractable, we propose a tractable approximation that allows us to scale our method to high-capacity neural network models. We demonstrate that our method can effectively optimize high-dimensional design problems in a variety of disciplines such as chemistry, biology, and materials engineering.

Spiking neural networks (SNNs) are biology-inspired artificial neural networks (ANNs) that comprise of spiking neurons to process asynchronous discrete signals. While more efficient in power consumption and inference speed on the neuromorphic hardware, SNNs are usually difficult to train directly from scratch with spikes due to the discreteness. As an alternative, many efforts have been devoted to converting conventional ANNs into SNNs by copying the weights from ANNs and adjusting the spiking threshold potential of neurons in SNNs. Researchers have designed new SNN architectures and conversion algorithms to diminish the conversion error. However, an effective conversion should address the difference between the SNN and ANN architectures with an efficient approximation of the loss function, which is missing in the field. In this work, we analyze the conversion error by recursive reduction to layer-wise summation and propose a novel strategic pipeline that transfers the weights to the target SNN by combining threshold balance and soft-reset mechanisms. This pipeline enables almost no accuracy loss between the converted SNNs and conventional ANNs with only $\sim1/10$ of the typical SNN simulation time. Our method is promising to get implanted onto embedded platforms with better support of SNNs with limited energy and memory. Codes are available at https://github.com/Jackn0/snn_optimal_conversion_pipeline.

Graphs are one of the most important data structures for representing pairwise relations between objects. Specifically, a graph embedded in a Euclidean space is essential to solving real problems, such as physical simulations. A crucial requirement for applying graphs in Euclidean spaces to physical simulations is learning and inferring the isometric transformation invariant and equivariant features in a computationally efficient manner. In this paper, we propose a set of transformation invariant and equivariant models based on graph convolutional networks, called IsoGCNs. We demonstrate that the proposed model has a competitive performance compared to state-of-the-art methods on tasks related to geometrical and physical simulation data. Moreover, the proposed model can scale up to graphs with 1M vertices and conduct an inference faster than a conventional finite element analysis, which the existing equivariant models cannot achieve.

Subspace clustering is an unsupervised clustering technique designed to cluster data that is supported on a union of linear subspaces, with each subspace defining a cluster with dimension lower than the ambient space. Many existing formulations for this problem are based on exploiting the self-expressive property of linear subspaces, where any point within a subspace can be represented as linear combination of other points within the subspace. To extend this approach to data supported on a union of non-linear manifolds, numerous studies have proposed learning an embedding of the original data using a neural network which is regularized by a self-expressive loss function on the data in the embedded space to encourage a union of linear subspaces prior on the data in the embedded space. Here we show that there are a number of potential flaws with this approach which have not been adequately addressed in prior work. In particular, we show the model formulation is often ill-posed in that it can lead to a degenerate embedding of the data, which need not correspond to a union of subspaces at all and is poorly suited for clustering. We validate our theoretical results experimentally and also repeat prior experiments reported in the literature, where we conclude that a significant portion of the previously claimed performance benefits can be attributed to an ad-hoc post processing step rather than the deep subspace clustering model.

Training on synthetic data can be beneficial for label or data-scarce scenarios. However, synthetically trained models often suffer from poor generalization in real domains due to domain gaps. In this work, we make a key observation that the diversity of the learned feature embeddings plays an important role in the generalization performance. To this end, we propose contrastive synthetic-to-real generalization (CSG), a novel framework that leverage the pre-trained ImageNet knowledge to prevent overfitting to the synthetic domain, while promoting the diversity of feature embeddings as an inductive bias to improve generalization. In addition, we enhance the proposed CSG framework with attentional pooling (A-pool) to let the model focus on semantically important regions and further improve its generalization. We demonstrate the effectiveness of CSG on various synthetic training tasks, exhibiting state-of-the-art performance on zero-shot domain generalization.

We propose a general, yet simple patch that can be applied to existing regularization-based continual learning methods called classifier-projection regularization (CPR). Inspired by both recent results on neural networks with wide local minima and information theory, CPR adds an additional regularization term that maximizes the entropy of a classifier's output probability. We demonstrate that this additional term can be interpreted as a projection of the conditional probability given by a classifier's output to the uniform distribution. By applying the Pythagorean theorem for KL divergence, we then prove that this projection may (in theory) improve the performance of continual learning methods. In our extensive experimental results, we apply CPR to several state-of-the-art regularization-based continual learning methods and benchmark performance on popular image recognition datasets. Our results demonstrate that CPR indeed promotes a wide local minima and significantly improves both accuracy and plasticity while simultaneously mitigating the catastrophic forgetting of baseline continual learning methods. The codes and scripts for this work are available at https://github.com/csm9493/CPR_CL.

We propose the challenge of rapid task-solving in novel environments (RTS), wherein an agent must solve a series of tasks as rapidly as possible in an unfamiliar environment. An effective RTS agent must balance between exploring the unfamiliar environment and solving its current task, all while building a model of the new environment over which it can plan when faced with later tasks. While modern deep RL agents exhibit some of these abilities in isolation, none are suitable for the full RTS challenge. To enable progress toward RTS, we introduce two challenge domains: (1) a minimal RTS challenge called the Memory&Planning Game and (2) One-Shot StreetLearn Navigation, which introduces scale and complexity from real-world data. We demonstrate that state-of-the-art deep RL agents fail at RTS in both domains, and that this failure is due to an inability to plan over gathered knowledge. We develop Episodic Planning Networks (EPNs) and show that deep-RL agents with EPNs excel at RTS, outperforming the nearest baseline by factors of 2-3 and learning to navigate held-out StreetLearn maps within a single episode. We show that EPNs learn to execute a value iteration-like planning algorithm and that they generalize to situations beyond their training experience.

The \textit{memorization effects} of deep networks show that they will first memorize training data with clean labels and then those with noisy labels. The \textit{early stopping} method therefore can be exploited for learning with noisy labels. However, the side effect brought by noisy labels will influence the memorization of clean labels before early stopping. In this paper, motivated by the \textit{lottery ticket hypothesis} which shows that only partial parameters are important for generalization, we find that only partial parameters are important for fitting clean labels and generalize well, which we term as \textit{critical parameters}; while the other parameters tend to fit noisy labels and cannot generalize well, which we term as \textit{non-critical parameters}. Based on this, we propose \textit{robust early-learning} to reduce the side effect of noisy labels before early stopping and thus enhance the memorization of clean labels. Specifically, in each iteration, we divide all parameters into the critical and non-critical ones, and then perform different update rules for different types of parameters. Extensive experiments on benchmark-simulated and real-world label-noise datasets demonstrate the superiority of the proposed method over the state-of-the-art label-noise learning methods.

While training can mostly be accelerated by reducing the time needed to propagate neural gradients (loss gradients with respect to the intermediate neural layer outputs) back throughout the model, most previous works focus on the quantization/pruning of weights and activations. These methods are often not applicable to neural gradients, which have very different statistical properties. Distinguished from weights and activations, we find that the distribution of neural gradients is approximately lognormal. Considering this, we suggest two closed-form analytical methods to reduce the computational and memory burdens of neural gradients. The first method optimizes the floating-point format and scale of the gradients. The second method accurately sets sparsity thresholds for gradient pruning. Each method achieves state-of-the-art results on ImageNet. To the best of our knowledge, this paper is the first to (1) quantize the gradients to 6-bit floating-point formats, or (2) achieve up to 85% gradient sparsity --- in each case without accuracy degradation. Reference implementation accompanies the paper in the supplementary material.

Hyperbolic spaces, which have the capacity to embed tree structures without distortion owing to their exponential volume growth, have recently been applied to machine learning to better capture the hierarchical nature of data. In this study, we generalize the fundamental components of neural networks in a single hyperbolic geometry model, namely, the Poincaré ball model. This novel methodology constructs a multinomial logistic regression, fully-connected layers, convolutional layers, and attention mechanisms under a unified mathematical interpretation, without increasing the parameters. Experiments show the superior parameter efficiency of our methods compared to conventional hyperbolic components, and stability and outperformance over their Euclidean counterparts.

A key challenge for reinforcement learning (RL) consists of learning in environments with sparse extrinsic rewards. In contrast to current RL methods, humans are able to learn new skills with little or no reward by using various forms of intrinsic motivation. We propose AMIGo, a novel agent incorporating -- as form of meta-learning -- a goal-generating teacher that proposes Adversarially Motivated Intrinsic Goals to train a goal-conditioned "student" policy in the absence of (or alongside) environment reward. Specifically, through a simple but effective "constructively adversarial" objective, the teacher learns to propose increasingly challenging -- yet achievable -- goals that allow the student to learn general skills for acting in a new environment, independent of the task to be solved. We show that our method generates a natural curriculum of self-proposed goals which ultimately allows the agent to solve challenging procedurally-generated tasks where other forms of intrinsic motivation and state-of-the-art RL methods fail.

We propose a novel task of joint few-shot recognition and novel-view synthesis: given only one or few images of a novel object from arbitrary views with only category annotation, we aim to simultaneously learn an object classifier and generate images of that type of object from new viewpoints. While existing work copes with two or more tasks mainly by multi-task learning of shareable feature representations, we take a different perspective. We focus on the interaction and cooperation between a generative model and a discriminative model, in a way that facilitates knowledge to flow across tasks in complementary directions. To this end, we propose bowtie networks that jointly learn 3D geometric and semantic representations with a feedback loop. Experimental evaluation on challenging fine-grained recognition datasets demonstrates that our synthesized images are realistic from multiple viewpoints and significantly improve recognition performance as ways of data augmentation, especially in the low-data regime.

A common approach for compressing Natural Language Processing (NLP) networks is to encode the embedding layer as a matrix $A\in\mathbb{R}^{n\times d}$, compute its rank-$j$ approximation $A_j$ via SVD (Singular Value Decomposition), and then factor $A_j$ into a pair of matrices that correspond to smaller fully-connected layers to replace the original embedding layer. Geometrically, the rows of $A$ represent points in $\mathbb{R}^d$, and the rows of $A_j$ represent their projections onto the $j$-dimensional subspace that minimizes the sum of squared distances (``errors'') to the points. In practice, these rows of $A$ may be spread around $k>1$ subspaces, so factoring $A$ based on a single subspace may lead to large errors that turn into large drops in accuracy. Inspired by \emph{projective clustering} from computational geometry, we suggest replacing this subspace by a set of $k$ subspaces, each of dimension $j$, that minimizes the sum of squared distances over every point (row in $A$) to its \emph{closest} subspace. Based on this approach, we provide a novel architecture that replaces the original embedding layer by a set of $k$ small layers that operate in parallel and are then recombined with a single fully-connected layer. Extensive experimental results on the GLUE benchmark yield networks that are both more accurate and smaller compared to the standard matrix factorization (SVD). For example, we further compress DistilBERT by reducing the size of the embedding layer by $40\%$ while incurring only a $0.5\%$ average drop in accuracy over all nine GLUE tasks, compared to a $2.8\%$ drop using the existing SVD approach. On RoBERTa we achieve $43\%$ compression of the embedding layer with less than a $0.8\%$ average drop in accuracy as compared to a $3\%$ drop previously.

We propose a simple and efficient multi-hop dense retrieval approach for answering complex open-domain questions, which achieves state-of-the-art performance on two multi-hop datasets, HotpotQA and multi-evidence FEVER. Contrary to previous work, our method does not require access to any corpus-specific information, such as inter-document hyperlinks or human-annotated entity markers, and can be applied to any unstructured text corpus. Our system also yields a much better efficiency-accuracy trade-off, matching the best published accuracy on HotpotQA while being 10 times faster at inference time.

Existing literature in Continual Learning (CL) has focused on overcoming catastrophic forgetting, the inability of the learner to recall how to perform tasks observed in the past. There are however other desirable properties of a CL system, such as the ability to transfer knowledge from previous tasks and to scale memory and compute sub-linearly with the number of tasks. Since most current benchmarks focus only on forgetting using short streams of tasks, we first propose a new suite of benchmarks to probe CL algorithms across these new axes. Finally, we introduce a new modular architecture, whose modules represent atomic skills that can be composed to perform a certain task. Learning a task reduces to figuring out which past modules to re-use, and which new modules to instantiate to solve the current task. Our learning algorithm leverages a task-driven prior over the exponential search space of all possible ways to combine modules, enabling efficient learning on long streams of tasks. Our experiments show that this modular architecture and learning algorithm perform competitively on widely used CL benchmarks while yielding superior performance on the more challenging benchmarks we introduce in this work. The Benchmark is publicly available at https://github.com/facebookresearch/CTrLBenchmark.