Deep generative frameworks including GANs and normalizing flow models have proven successful at filling in missing values in partially observed data samples by effectively learning -- either explicitly or implicitly -- complex, high-dimensional statistical distributions. In tasks where the data available for learning is only partially observed, however, their performance decays monotonically as a function of the data missingness rate. In high missing data rate regimes (e.g., 60% and above), it has been observed that state-of-the-art models tend to break down and produce unrealistic and/or semantically inaccurate data. We propose a novel framework to facilitate the learning of data distributions in high paucity scenarios that is inspired by traditional formulations of solutions to ill-posed problems. The proposed framework naturally stems from posing the process of learning from incomplete data as a joint optimization task of the parameters of the model being learned and the missing data values. The method involves enforcing a prior regularization term that seamlessly integrates with objectives used to train explicit and tractable deep generative frameworks such as deep normalizing flow models. We demonstrate via extensive experimental validation that the proposed framework outperforms competing techniques, particularly as the rate of data paucity approaches unity.
Institute of Information Science, Academia Sinica（中央研究院资讯科学研究所）；Department of Computer Science and Engineering, Texas A&M University（德州农工大学计算机科学与工程系）
In this paper, we propose an approach to effectively accelerating the computation of continuous normalizing flow (CNF), which has been proven to be a powerful tool for the tasks such as variational inference and density estimation. The training time cost of CNF can be extremely high because the required number of function evaluations (NFE) for solving corresponding ordinary differential equations (ODE) is very large. We think that the high NFE results from large truncation errors of solving ODEs. To address the problem, we propose to add a regularization. The regularization penalizes the diffe…
Department of Electrical and Computer Engineering（电气与计算机工程系）；Duke University（杜克大学）
We introduce Projected Latent Markov Chain Monte Carlo (PL-MCMC), a technique for sampling from the high-dimensional conditional distributions learned by a normalizing flow. We prove that a Metropolis-Hastings implementation of PL-MCMC asymptotically samples from the exact conditional distributions associated with a normalizing flow. As a conditional sampling method, PL-MCMC enables Monte Carlo Expectation Maximization (MC-EM) training of normalizing flows from incomplete data. Through experimental tests applying normalizing flows to missing data tasks for a variety of data sets, we demonstrat…
Catastrophic forgetting (CF) happens whenever a neural network overwrites past knowledge while being trained on new tasks. Common techniques to handle CF include regularization of the weights (using, e.g., their importance on past tasks), and rehearsal strategies, where the network is constantly re-trained on past data. Generative models have also been applied for the latter, in order to have endless sources of data. In this paper, we propose a novel method that combines the strengths of regularization and generative-based rehearsal approaches. Our generative model consists of a normalizing fl…
Generating a Boltzmann distribution in high dimension has recently been achieved with Normalizing Flows, which enable fast and exact computation of the generated density, and thus unbiased estimation of expectations. However, current implementations rely on accurate training data, which typically comes from computationally expensive simulations. There is therefore a clear incentive to train models with incomplete or no data by relying solely on the target density, which can be obtained from a physical energy model (up to a constant factor). For that purpose, we analyze the properties of standa…
Georgia Institute of Technology（佐治亚理工学院）；Anonymous University（匿名大学）
Incomplete instances with various missing attributes in many real-world applications have brought challenges to the classification tasks. Missing values imputation methods are often employed to replace the missing values with substitute values. However, this process often separates the imputation and classification, which may lead to inferior performance since label information are often ignored during imputation. Moreover, traditional methods may rely on improper assumptions to initialize the missing values, whereas the unreliability of such initialization might lead to inferior performance. …
University of Oxford（牛津大学）；University of Bath（巴斯大学）
Training Neural Ordinary Differential Equations (ODEs) is often computationally expensive. Indeed, computing the forward pass of such models involves solving an ODE which can become arbitrarily complex during training. Recent works have shown that regularizing the dynamics of the ODE can partially alleviate this. In this paper we propose a new regularization technique: randomly sampling the end time of the ODE during training. The proposed regularization is simple to implement, has negligible overhead and is effective across a wide variety of tasks. Further, the technique is orthogonal to seve…
Efficient Regression-Based Training of Normalizing Flows for Boltzmann Generators
Mila – Québec AI Institute（魁北克AI研究所）；Université de Montréal（蒙特利尔大学）；Massachusetts Institute of Technology（麻省理工学院）；University of Oxford（牛津大学）；AITHYRA；HEC Montréal（蒙特利尔HEC）
Normalizing flows, autoregressive models, variational autoencoders (VAEs), and deep energy-based models are among competing likelihood-based frameworks for deep generative learning. Among them, VAEs have the advantage of fast and tractable sampling and easy-to-access encoding networks. However, they are currently outperformed by other models such as normalizing flows and autoregressive models. While the majority of the research in VAEs is focused on the statistical challenges, we explore the orthogonal direction of carefully designing neural architectures for hierarchical VAEs. We propose Nouv…
Rochester Data Science Consortium（罗切斯特数据科学联盟）；University of Rochester（罗切斯特大学）
We consider the topic of data imputation, a foundational task in machine learning that addresses issues with missing data. To that end, we propose MCFlow, a deep framework for imputation that leverages normalizing flow generative models and Monte Carlo sampling. We address the causality dilemma that arises when training models with incomplete data by introducing an iterative learning scheme which alternately updates the density estimate and the values of the missing entries in the training data. We provide extensive empirical validation of the effectiveness of the proposed method on standard m…
Flow-based conditional cardiac anatomy generation for virtual cohorts
Delft University of Technology（代尔夫特理工大学）；Hospital of the University of Pennsylvania（宾夕法尼亚大学医院）；Ghent University（根特大学）；University of Pennsylvania（宾夕法尼亚大学）
Max Planck Institute for Informatics（马克斯·普朗克信息学研究所）；Bosch Center for Artificial Intelligence（博世人工智能中心）；CISPA Helmholtz Center for Information Security（CISPA亥姆霍兹信息安全中心）
Prediction of future states of the environment and interacting agents is a key competence required for autonomous agents to operate successfully in the real world. Prior work for structured sequence prediction based on latent variable models imposes a uni-modal standard Gaussian prior on the latent variables. This induces a strong model bias which makes it challenging to fully capture the multi-modality of the distribution of the future states. In this work, we introduce Conditional Flow Variational Autoencoders (CF-VAE) using our novel conditional normalizing flow based prior to capture compl…
We reinterpreting the variational inference in a new perspective. Via this way, we can easily prove that EM algorithm, VAE, GAN, AAE, ALI(BiGAN) are all special cases of variational inference. The proof also reveals the loss of standard GAN is incomplete and it explains why we need to train GAN cautiously. From that, we find out a regularization term to improve stability of GAN training.
We discuss an approach to probabilistic forecasting based on two chained machine-learning steps: a dimensional reduction step that learns a reduction map of predictor information to a low-dimensional space in a manner designed to preserve information about forecast quantities; and a density estimation step that uses the probabilistic machine learning technique of normalizing flows to compute the joint probability density of reduced predictors and forecast quantities. This joint density is then renormalized to produce the conditional forecast distribution. In this method, probabilistic calibrat…
Department of Computer Science and Engineering, Washington University in St. Louis（圣路易斯华盛顿大学计算机科学与工程系）；Department of Electrical and Systems Engineering, Washington University in St. Louis（圣路易斯华盛顿大学电气与系统工程系）；Department of Bioengineering, University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校生物工程系）
Medical image reconstruction is typically an ill-posed inverse problem. In order to address such ill-posed problems, the prior distribution of the sought after object property is usually incorporated by means of some sparsity-promoting regularization. Recently, prior distributions for images estimated using generative adversarial networks (GANs) have shown great promise in regularizing some of these image reconstruction problems. In this work, we apply an image-adaptive GAN-based reconstruction method (IAGAN) to reconstruct high fidelity images from incomplete medical imaging data. It is obser…
Philips Co-Creation Center（飞利浦共创中心）；University of Tsukuba（筑波大学）；I Dragon Corporation（I Dragon公司）；The Tokyo Foundation for Policy Research（东京政策研究基金会）；Lowe’s Services India Pvt. Ltd.（劳氏印度服务私人有限公司）
Virtual Adversarial Training (VAT) has shown impressive results among recently developed regularization methods called consistency regularization. VAT utilizes adversarial samples, generated by injecting perturbation in the input space, for training and thereby enhances the generalization ability of a classifier. However, such adversarial samples can be generated only within a very small area around the input data point, which limits the adversarial effectiveness of such samples. To address this problem we propose LVAT (Latent space VAT), which injects perturbation in the latent space instead …
Inverse problems consist in reconstructing signals from incomplete sets of measurements and their performance is highly dependent on the quality of the prior knowledge encoded via regularization. While traditional approaches focus on obtaining a unique solution, an emerging trend considers exploring multiple feasibile solutions. In this paper, we propose a method to generate multiple reconstructions that fit both the measurements and a data-driven prior learned by a generative adversarial network. In particular, we show that, starting from an initial solution, it is possible to find directions…
Offline reinforcement learning aims to train a policy on a pre-recorded and fixed dataset without any additional environment interactions. There are two major challenges in this setting: (1) extrapolation error caused by approximating the value of state-action pairs not well-covered by the training data and (2) distributional shift between behavior and inference policies. One way to tackle these problems is to induce conservatism - i.e., keeping the learned policies closer to the behavioral ones. To achieve this, we build upon recent works on learning policies in latent action spaces and use a…
Riemannian AmbientFlow: Towards Simultaneous Manifold Learning and Generative Modeling from Corrupted Data
Department of Mathematics, University of California, Los Angeles（数学系，加州大学洛杉矶分校）；Department of Statistics and Data Science, University of California, Los Angeles（统计与数据科学系，加州大学洛杉矶分校）
An Actor-Critic Framework for Continuous-Time Jump-Diffusion Controls with Normalizing Flows
Yau Mathematical Sciences Center, Tsinghua University（清华大学丘成桐数学科学中心）；Department of Mathematics, Tsinghua University（清华大学数学系）；Department of Mathematics, University of California, Santa Barbara（加州大学圣塔芭芭拉分校数学系）；Department of Statistics and Applied Probability, University of California, Santa Barbara（加州大学圣塔芭芭拉分校统计与应用概率系）；Yanqi Lake Beijing Institute of Mathematical Sciences and Applications（北京雁栖湖应用数学研究院）
LIT AI Lab（LIT人工智能实验室）；Institute of Bioinformatics（生物信息学研究所）；Johannes Kepler University Linz（约翰内斯·开普勒大学林茨分校）
Deep Learning has revolutionized vision via convolutional neural networks (CNNs) and natural language processing via recurrent neural networks (RNNs). However, success stories of Deep Learning with standard feed-forward neural networks (FNNs) are rare. FNNs that perform well are typically shallow and, therefore cannot exploit many levels of abstract representations. We introduce self-normalizing neural networks (SNNs) to enable high-level abstract representations. While batch normalization requires explicit normalization, neuron activations of SNNs automatically converge towards zero mean and …
Multi-task learning is a type of transfer learning that trains multiple tasks simultaneously and leverages the shared information between related tasks to improve the generalization performance. However, missing features in the input matrix is a much more difficult problem which needs to be carefully addressed. Removing records with missing values can significantly reduce the sample size, which is impractical for datasets with large percentage of missing values. Popular imputation methods often distort the covariance structure of the data, which causes inaccurate inference. In this paper we pr…
Batch normalization is widely used in deep learning to normalize intermediate activations. Deep networks suffer from notoriously increased training complexity, mandating careful initialization of weights, requiring lower learning rates, etc. These issues have been addressed by Batch Normalization (\textbf{BN}), by normalizing the inputs of activations to zero mean and unit standard deviation. Making this batch normalization part of the training process dramatically accelerates the training process of very deep networks. A new field of research has been going on to examine the exact theoretical…
Manifold-based Incomplete Multi-view Clustering via Bi-Consistency Guidance
College of Information Science and Technology, Dalian Maritime University（大连海事大学信息科学技术学院）；University of Technology, Sydney（悉尼科技大学）；Hefei University of Technology（合肥工业大学）
tPARAFAC2: Tracking evolving patterns in (incomplete) temporal data
Simula Metropolitan Center for Digital Engineering（Simula 数字工程中心）；Faculty of Technology, Art and Design, OsloMet（技术、艺术与设计学院）；Institute for Numerical and Applied Mathematics, Georg-August-Universität Göttingen（数值与应用数学研究所）
Most practical data science problems encounter missing data. A wide variety of solutions exist, each with strengths and weaknesses that depend upon the missingness-generating process. Here we develop a theoretical framework for training and inference using only observed variables enabling modeling of incomplete datasets without imputation. Using an information and measure-theoretic argument we construct models with latent space representations that regularize against the potential bias introduced by missing data. The theoretical properties of this approach are demonstrated empirically using a …
New York University（纽约大学）；DeepMind（深度思维）；Google Research（谷歌研究院）
Learning new tasks continuously without forgetting on a constantly changing data distribution is essential for real-world problems but extremely challenging for modern deep learning. In this work we propose HCL, a Hybrid generative-discriminative approach to Continual Learning for classification. We model the distribution of each task and each class with a normalizing flow. The flow is used to learn the data distribution, perform classification, identify task changes, and avoid forgetting, all leveraging the invertibility and exact likelihood which are uniquely enabled by the normalizing flow …
University of Illinois at Chicago（伊利诺伊大学芝加哥分校）；Shenzhen University（深圳大学）
In the era of big data, it is common to have data with multiple modalities or coming from multiple sources, known as "multi-view data". Multi-view clustering provides a natural way to generate clusters from such data. Since different views share some consistency and complementary information, previous works on multi-view clustering mainly focus on how to combine various numbers of views to improve clustering performance. However, in reality, each view may be incomplete, i.e., instances missing in the view. Furthermore, the size of data could be extremely huge. It is unrealistic to apply multi-…
Carnegie Mellon University（卡内基梅隆大学）；RIKEN Center for Artificial Intelligence Project（理化学研究所人工智能项目中心）
There has been an increased interest in multimodal language processing including multimodal dialog, question answering, sentiment analysis, and speech recognition. However, naturally occurring multimodal data is often imperfect as a result of imperfect modalities, missing entries or noise corruption. To address these concerns, we present a regularization method based on tensor rank minimization. Our method is based on the observation that high-dimensional multimodal time series data often exhibit correlations across time and modalities which leads to low-rank tensor representations. However, t…
Preparatory Institute of Engineer of Faculty of science of Monastir（莫纳斯提尔科学学院工程师预备学院）；Faculty of science of Monastir（莫纳斯提尔科学学院）
Bayesian networks (BN) are used in a big range of applications but they have one issue concerning parameter learning. In real application, training data are always incomplete or some nodes are hidden. To deal with this problem many learning parameter algorithms are suggested foreground EM, Gibbs sampling and RBE algorithms. In order to limit the search space and escape from local maxima produced by executing EM algorithm, this paper presents a learning parameter algorithm that is a fusion of EM and RBE algorithms. This algorithm incorporates the range of a parameter into the EM algorithm. This…
Vanderbilt University（范德堡大学）；Vanderbilt University Medical Center（范德堡大学医学中心）
Data from multi-modality provide complementary information in clinical prediction, but missing data in clinical cohorts limits the number of subjects in multi-modal learning context. Multi-modal missing imputation is challenging with existing methods when 1) the missing data span across heterogeneous modalities (e.g., image vs. non-image); or 2) one modality is largely missing. In this paper, we address imputation of missing data by modeling the joint distribution of multi-modal data. Motivated by partial bidirectional generative adversarial net (PBiGAN), we propose a new Conditional PBiGAN (C…
Catastrophic forgetting (CF) happens whenever a neural network overwrites past knowledge while being trained on new tasks. Common techniques to handle CF include regularization of the weights (using, e.g., their importance on past tasks), and rehearsal strategies, where the network is constantly re-trained on past data. Generative models have also been applied for the latter, in order to have endless sources of data. In this paper, we propose a novel method that combines the strengths of regularization and generative-based rehearsal approaches. Our generative model consists of a normalizing fl…
Learning Nonlinear Factor Models with Unknown Monotone Links from Incomplete and Noisy Data
School of Computation, Information and Technology, Technical University of Munich, Germany（计算、信息与技术学院，慕尼黑技术大学，德国）；Department of Economics, Virginia Tech, USA（经济系，弗吉尼亚理工学院，美国）；Munich Institute of Robotics and Machine Intelligence（慕尼黑机器人与人工智能研究所）
University of Electronic Science and Technology of China（电子科技大学）
Tensor completion is a challenging problem with various applications. Many related models based on the low-rank prior of the tensor have been proposed. However, the low-rank prior may not be enough to recover the original tensor from the observed incomplete tensor. In this paper, we prose a tensor completion method by exploiting both the low-rank and sparse prior of tensor. Specifically, the tensor completion task can be formulated as a low-rank minimization problem with a sparse regularizer. The low-rank property is depicted by the tensor truncated nuclear norm based on tensor singular value …
Department of Science Education（科学教育系）；Center for Theoretical Physics and Artificial Intelligence Institute（理论物理与人工智能研究所）；School of Computational Sciences（计算科学学院）
FedMAC: Tackling Partial-Modality Missing in Federated Learning with Cross-Modal Aggregation and Contrastive Regularization
Hanoi University of Science and Technology（河内科学技术大学）；Nagoya University（名古屋大学）；Washington State University（华盛顿州立大学）；Swiss Federal Institute of Technology Lausanne（洛桑联邦理工学院）
