ICML 2020 论文列表

We present a new method for separating a mixed audio sequence, in which multiple voices speak simultaneously. The new method employs gated neural networks that are trained to separate the voices at multiple processing steps, while maintaining the speaker in each output channel fixed. A different model is trained for every number of possible speakers, and the model with the largest number of speakers is employed to select the actual number of speakers in a given sample. Our method greatly outperforms the current state of the art, which, as we show, is not competitive for more than two speakers.

Humans intuitively recognize objects’ physical properties and predict their motion, even when the objects are engaged in complicated interactions. The abilities to perform physical reasoning and to adapt to new environments, while intrinsic to humans, remain challenging to state-of-the-art computational models. In this work, we present a neural model that simultaneously reasons about physics and makes future predictions based on visual and dynamics priors. The visual prior predicts a particle-based representation of the system from visual observations. An inference module operates on those particles, predicting and refining estimates of particle locations, object states, and physical parameters, subject to the constraints imposed by the dynamics prior, which we refer to as visual grounding. We demonstrate the effectiveness of our method in environments involving rigid objects, deformable materials, and fluids. Experiments show that our model can infer the physical properties within a few observations, which allows the model to quickly adapt to unseen scenarios and make accurate predictions into the future.

Deep Neural Networks (DNNs) achieve the state-of-the-art results on a wide range of image processing tasks, however, the majority of such solutions are problem-specific, like most AI algorithms. The One Network to Solve Them All (OneNet) procedure has been suggested to resolve this issue by exploiting a DNN as the proximal operator in Alternating Direction Method of Multipliers (ADMM) solvers for various imaging problems. In this work, we make two contributions, both facilitating end-to-end learning using backpropagation. First, we generalize OneNet to videos by augmenting its convolutional prior network with bidirectional recurrent connections; second, we extend the fixed fully connected linear ADMM data step with another trainable bidirectional convolutional recurrent network. In our computational experiments on the Rotated MNIST, Scanned CIFAR-10 and UCF-101 data sets, the proposed modifications improve performance by a large margin compared to end-to-end convolutional OneNet and 3D Wavelet sparsity on several video processing problems: pixelwise inpainting-denoising, blockwise inpainting, scattered inpainting, super resolution, compressive sensing, deblurring, frame interpolation, frame prediction and colorization. Our two contributions are complementary, and using them together yields the best results.

In video prediction tasks, one major challenge is to capture the multi-modal nature of future contents and dynamics. In this work, we propose a simple yet effective framework that can efficiently predict plausible future states, where the key insight is that the potential distribution of a sequence could be approximated with analogous ones in a repertoire of training pool, namely, expert examples. By further incorporating a novel optimization scheme into the training procedure, plausible predictions can be sampled efficiently from distribution constructed from the retrieved examples. Meanwhile, our method could be seamlessly integrated with existing stochastic predictive models; significant enhancement is observed with comprehensive experiments in both quantitative and qualitative aspects. We also demonstrate the generalization ability to predict the motion of unseen class, i.e., without access to corresponding data during training phase. Project Page: \hyperlink{https://sites.google.com/view/vpeg-supp/home.}{https://sites.google.com/view/vpeg-supp/home.}

Label distribution covers a certain number of labels, representing the degree to which each label describes the instance. When dealing with label ambiguity, label distribution could describe the supervised information in a fine-grained way. Unfortunately, many training sets only contain simple logical labels rather than label distributions due to the difficulty of obtaining label distributions directly. To solve this problem, we consider the label distributions as the latent vectors and infer them from the logical labels in the training datasets by using variational inference. After that, we induce a predictive model to train the label distribution data by employing the multi-output regression technique. The recovery experiment on thirteen real-world LDL datasets and the predictive experiment on ten multi-label learning datasets validate the advantage of our approach over the state-of-the-art approaches.

The recent development of flexible and scalable variational inference algorithms has popularized the use of deep probabilistic models in a wide range of applications. However, learning and reasoning about high-dimensional models with nondifferentiable densities are still a challenge. For such a model, inference algorithms struggle to estimate the gradients of variational objectives accurately, due to high variance in their estimates. To tackle this challenge, we present a novel variational inference algorithm for sequential data, which performs well even when the density from the model is not differentiable, for instance, due to the use of discrete random variables. The key feature of our algorithm is that it estimates future likelihoods at all time steps. The estimated future likelihoods form the core of our new low-variance gradient estimator. We formally analyze our gradient estimator from the perspective of variational objective, and show the effectiveness of our algorithm with synthetic and real datasets.

Learning from demonstrations can be challenging when the quality of demonstrations is diverse, and even more so when the quality is unknown and there is no additional information to estimate the quality. We propose a new method for imitation learning in such scenarios. We show that simple quality-estimation approaches might fail due to compounding error, and fix this issue by jointly estimating both the quality and reward using a variational approach. Our method is easy to implement within reinforcement-learning frameworks and also achieves state-of-the-art performance on continuous-control benchmarks.Our work enables scalable and data-efficient imitation learning under more realistic settings than before.

We propose a novel algorithm for quantizing continuous latent representations in trained models. Our approach applies to deep probabilistic models, such as variational autoencoders (VAEs), and enables both data and model compression. Unlike current end-to-end neural compression methods that cater the model to a fixed quantization scheme, our algorithm separates model design and training from quantization. Consequently, our algorithm enables “plug-and-play” compression with variable rate-distortion trade-off, using a single trained model. Our algorithm can be seen as a novel extension of arithmetic coding to the continuous domain, and uses adaptive quantization accuracy based on estimates of posterior uncertainty. Our experimental results demonstrate the importance of taking into account posterior uncertainties, and show that image compression with the proposed algorithm outperforms JPEG over a wide range of bit rates using only a single standard VAE. Further experiments on Bayesian neural word embeddings demonstrate the versatility of the proposed method.

Variational Autoencoders (VAEs) represent the given data in a low-dimensional latent space, which is generally assumed to be Euclidean. This assumption naturally leads to the common choice of a standard Gaussian prior over continuous latent variables. Recent work has, however, shown that this prior has a detrimental effect on model capacity, leading to subpar performance. We propose that the Euclidean assumption lies at the heart of this failure mode. To counter this, we assume a Riemannian structure over the latent space, which constitutes a more principled geometric view of the latent codes, and replace the standard Gaussian prior with a Riemannian Brownian motion prior. We propose an efficient inference scheme that does not rely on the unknown normalizing factor of this prior. Finally, we demonstrate that this prior significantly increases model capacity using only one additional scalar parameter.

Stochastic particle-optimization sampling (SPOS) is a recently-developed scalable Bayesian sampling framework unifying stochastic gradient MCMC (SG-MCMC) and Stein variational gradient descent (SVGD) algorithms based on Wasserstein gradient flows. With a rigorous non-asymptotic convergence theory developed, SPOS can avoid the particle-collapsing pitfall of SVGD. However, the variance-reduction effect in SPOS has not been clear. In this paper, we address this gap by presenting several variance-reduction techniques for SPOS. Specifically, we propose three variants of variance-reduced SPOS, called SAGA particle-optimization sampling (SAGA-POS), SVRG particle-optimization sampling (SVRG-POS) and a variant of SVRG-POS which avoids full gradient computations, denoted as SVRG-POS$^+$. Importantly, we provide non-asymptotic convergence guarantees for these algorithms in terms of the 2-Wasserstein metric and analyze their complexities. The results show our algorithms yield better convergence rates than existing variance-reduced variants of stochastic Langevin dynamics, though more space is required to store the particles in training. Our theory aligns well with experimental results on both synthetic and real datasets.

Particle-based Variational Inference methods (ParVIs), like Stein Variational Gradient Descent, are nonparametric variational inference methods that optimize a set of particles to best approximate a target distribution. ParVIs have been proposed as efficient approximate inference algorithms and as potential alternatives to MCMC methods. However, to our knowledge, the quality of the posterior approximation of particles from ParVIs has not been examined before for large-scale Bayesian inference problems. We conduct this analysis and evaluate the sample quality of particles produced by ParVIs, and we find that existing ParVI approaches using stochastic gradients converge insufficiently fast under sample quality metrics. We propose a novel variance reduction and quasi-Newton preconditioning framework for ParVIs, by leveraging the Riemannian structure of the Wasserstein space and advanced Riemannian optimization algorithms. Experimental results demonstrate the accelerated convergence of variance reduction and quasi-Newton methods for ParVIs for accurate posterior inference in large-scale and ill-conditioned problems.

We propose an accelerated version of stochastic variance reduced coordinate descent – ASVRCD. As other variance reduced coordinate descent methods such as SEGA or SVRCD, our method can deal with problems that include a non-separable and non-smooth regularizer, while accessing a random block of partial derivatives in each iteration only. However, ASVRCD incorporates Nesterov’s momentum, which offers favorable iteration complexity guarantees over both SEGA and SVRCD. As a by-product of our theory, we show that a variant of Katyusha (Allen-Zhu, 2017) is a specific case of ASVRCD, recovering the optimal oracle complexity for the finite sum objective.

Deep autoregressive models compute point likelihood estimates of individual data points. However, many applications (i.e., database cardinality estimation), require estimating range densities, a capability that is under-explored by current neural density estimation literature. In these applications, fast and accurate range density estimates over high-dimensional data directly impact user-perceived performance. In this paper, we explore a technique for accelerating range density estimation over deep autoregressive models. This technique, called variable skipping, exploits the sparse structure of range density queries to avoid sampling unnecessary variables during approximate inference. We show that variable skipping provides 10-100x efficiency improvements when targeting challenging high-quantile error metrics, enables complex applications such as text pattern matching, and can be realized via a simple data augmentation procedure without changing the usual maximum likelihood objective.

Generative flows are promising tractable models for density modeling that define probabilistic distributions with invertible transformations. However, tractability imposes architectural constraints on generative flows. In this work, we study a previously overlooked constraint that all the intermediate representations must have the same dimensionality with the data due to invertibility, limiting the width of the network. We propose VFlow to tackle this constraint on dimensionality. VFlow augments the data with extra dimensions and defines a maximum evidence lower bound (ELBO) objective for estimating the distribution of augmented data jointly with the variational data augmentation distribution. Under mild assumptions, we show that the maximum ELBO solution of VFlow is always better than the original maximum likelihood solution. For image density modeling on the CIFAR-10 dataset, VFlow achieves a new state-of-the-art 2.98 bits per dimension.

Sparse principal component analysis (PCA) is a widely-used dimensionality reduction tool in statistics and machine learning. Most methods mentioned in literature are either heuristics for good primal feasible solutions under statistical assumptions or ADMM-type algorithms with stationary/critical points convergence property for the regularized reformulation of sparse PCA. However, none of these methods can efficiently verify the quality of the solutions via comparing current objective values with their dual bounds, especially in model-free case. We propose a new framework that finds out upper (dual) bounds for the sparse PCA within polynomial time via solving a convex integer program (IP). We show that, in the worst-case, the dual bounds provided by the convex IP is within an affine function of the global optimal value. Moreover, in contrast to the semi-definition relaxation, this framework is much easier to scale on large cases. Numerical results on both artificial and real cases are reported to demonstrate the advantages of our method.

When quantizing neural networks, assigning each floating-point weight to its nearest fixed-point value is the predominant approach. We find that, perhaps surprisingly, this is not the best we can do. In this paper, we propose AdaRound, a better weight-rounding mechanism for post-training quantization that adapts to the data and the task loss. AdaRound is fast, does not require fine-tuning of the network, and only uses a small amount of unlabelled data. We start by theoretically analyzing the rounding problem for a pre-trained neural network. By approximating the task loss with a Taylor series expansion, the rounding task is posed as a quadratic unconstrained binary optimization problem. We simplify this to a layer-wise local loss and propose to optimize this loss with a soft relaxation. AdaRound not only outperforms rounding-to-nearest by a significant margin but also establishes a new state-of-the-art for post-training quantization on several networks and tasks. Without fine-tuning, we can quantize the weights of Resnet18 and Resnet50 to 4 bits while staying within an accuracy loss of 1%.

This paper explores a new research problem of unsupervised transfer learning across multiple spatiotemporal prediction tasks. Unlike most existing transfer learning methods that focus on fixing the discrepancy between supervised tasks, we study how to transfer knowledge from a zoo of unsupervisedly learned models towards another predictive network. Our motivation is that models from different sources are expected to understand the complex spatiotemporal dynamics from different perspectives, thereby effectively supplementing the new task, even if the task has sufficient training samples. Technically, we propose a differentiable framework named transferable memory. It adaptively distills knowledge from a bank of memory states of multiple pretrained RNNs, and applies it to the target network via a novel recurrent structure called the Transferable Memory Unit (TMU). Compared with finetuning, our approach yields significant improvements on three benchmarks for spatiotemporal prediction, and benefits the target task even from less relevant pretext ones.

Speech information can be roughly decomposed into four components: language content, timbre, pitch, and rhythm. Obtaining disentangled representations of these components is useful in many speech analysis and generation applications. Recently, state-of-the-art voice conversion systems have led to speech representations that can disentangle speaker-dependent and independent information. However, these systems can only disentangle timbre, while information about pitch, rhythm and content is still mixed together. Further disentangling the remaining speech components is an under-determined problem in the absence of explicit annotations for each component, which are difficult and expensive to obtain. In this paper, we propose SpeechSplit, which can blindly decompose speech into its four components by introducing three carefully designed information bottlenecks. SpeechSplit is among the first algorithms that can separately perform style transfer on timbre, pitch and rhythm without text labels. Our code is publicly available at https://github.com/auspicious3000/SpeechSplit.

The latent spaces of GAN models often have semantically meaningful directions. Moving in these directions corresponds to human-interpretable image transformations, such as zooming or recoloring, enabling a more controllable generation process. However, the discovery of such directions is currently performed in a supervised manner, requiring human labels, pretrained models, or some form of self-supervision. These requirements severely restrict a range of directions existing approaches can discover. In this paper, we introduce an unsupervised method to identify interpretable directions in the latent space of a pretrained GAN model. By a simple model-agnostic procedure, we find directions corresponding to sensible semantic manipulations without any form of (self-)supervision. Furthermore, we reveal several non-trivial findings, which would be difficult to obtain by existing methods, e.g., a direction corresponding to background removal. As an immediate practical benefit of our work, we show how to exploit this finding to achieve competitive performance for weakly-supervised saliency detection. The implementation of our method is available online.