Machine Learning Researcher, D. E. Shaw & Co

Incoming Assistant Professor (Fall 2021)

Department of Computer Science

University of Illinois at Urbana-Champaign

Email: hanzhao [AT] illinois (DOT) edu

[Curriculum Vitae] [Google Scholar] [Research Statement] [Thesis]

I obtained my Ph.D. from the Machine Learning Department, Carnegie Mellon University, where I was advised by Geoff Gordon. Previously, I obtained my BEng degree from the Computer Science Department at Tsinghua University and MMath from the University of Waterloo. I have a broad interest in both the theoretical and applied side of machine learning. In particular, I work on representation learning, probabilistic circuits, transfer/multitask learning, and their applications in natural language, signal processing and quantitative finance. I am currently a machine learning researcher at D. E. Shaw & Co. I am also an incoming tenure-track assistant professor at the Department of Computer Science, the University of Illinois at Urbana-Champaign. I am actively looking for self-motivated students starting Fall 2021. If you are interested in working with me, please feel free to send me an email with your CV. |

J. Chi, Y. Tian, G. Gordon and

In Proceedings of the 38th International Conference on Machine Learning (

[abs] [pdf] [code]

With the widespread deployment of large-scale prediction systems in high-stakes domains, e.g., face recognition, criminal justice, etc., disparity on prediction accuracy between different demographic subgroups has called for fundamental understanding on the source of such disparity and algorithmic intervention to mitigate it. In this paper, we study the accuracy disparity problem in regression. To begin with, we first propose an error decomposition theorem, which decomposes the accuracy disparity into the distance between marginal label distributions and the distance between conditional representations, to help explain why such accuracy disparity appears in practice. Motivated by this error decomposition and the general idea of distribution alignment with statistical distances, we then propose an algorithm to reduce this disparity, and analyze its game-theoretic optima of the proposed objective functions. To corroborate our theoretical findings, we also conduct experiments on five benchmark datasets. The experimental results suggest that our proposed algorithms can effectively mitigate accuracy disparity while maintaining the predictive power of the regression models.

H. Wang,

In Proceedings of the 38th International Conference on Machine Learning (

[abs] [pdf] [code]

Multi-task learning (MTL) aims to improve the generalization of several related tasks by learning them jointly. As a comparison, in addition to the joint training scheme, modern meta-learning allows unseen tasks with limited labels during the test phase, in the hope of fast adaptation over them. Despite the subtle difference between MTL and meta-learning in the problem formulation, both learning paradigms share the same insight that the shared structure between existing training tasks could lead to better generalization and adaptation. In this paper, we take one important step further to understand the close connection between these two learning paradigms, through both theoretical analysis and empirical investigation. Theoretically, we first demonstrate that MTL shares the same optimization formulation with a class of gradient-based meta-learning (GBML) algorithms. We then prove that for over-parameterized neural networks with sufficient depth, the learned predictive functions of MTL and GBML are close. In particular, this result implies that the predictions given by these two models are similar over the same unseen task. Empirically, we corroborate our theoretical findings by showing that, with proper implementation, MTL is competitive against state-of-the-art GBML algorithms on a set of few-shot image classification benchmarks. Since existing GBML algorithms often involve costly second-order bi-level optimization, our first-order MTL method is an order of magnitude faster on large-scale datasets such as mini-ImageNet. We believe this work could help bridge the gap between these two learning paradigms, and provide a computationally efficient alternative to GBML that also supports fast task adaptation.

P. Liao,

In Proceedings of the 38th International Conference on Machine Learning (

[abs] [pdf] [code]

While the advent of Graph Neural Networks (GNNs) has greatly improved node and graph representation learning in many applications, the neighborhood aggregation scheme exposes additional vulnerabilities to adversaries seeking to extract node-level information about sensitive attributes. In this paper, we study the problem of protecting sensitive attributes by information obfuscation when learning with graph structured data. We propose a framework to locally filter out pre-determined sensitive attributes via adversarial training with the total variation and the Wasserstein distance. Our method creates a strong defense against inference attacks, while only suffering small loss in task performance. Theoretically, we analyze the effectiveness of our framework against a worst-case adversary, and characterize an inherent trade-off between maximizing predictive accuracy and minimizing information leakage. Experiments across multiple datasets from recommender systems, knowledge graphs and quantum chemistry demonstrate that the proposed approach provides a robust defense across various graph structures and tasks, while producing competitive GNN encoders for downstream tasks.

B. Li, Y. Wang, S. Zhang, D. Li, T. Darrell, K. Keutzer and

In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (

[abs] [pdf] [code]

The success of supervised learning hinges on the assumption that the training and test data come from the same underlying distribution, which is often not valid in practice due to potential distribution shift. In light of this, most existing methods for unsupervised domain adaptation focus on achieving domain-invariant representations and small source domain error. However, recent works have shown that this is not sufficient to guarantee good generalization on the target domain, and in fact, is provably detrimental under label distribution shift. Furthermore, in many real-world applications it is often feasible to obtain a small amount of labeled data from the target domain and use them to facilitate model training with source data. Inspired by the above observations, in this paper we propose the first method that aims to simultaneously learn invariant representations and risks under the setting of semi-supervised domain adaptation (Semi-DA). First, we provide a finite sample bound for both classification and regression problems under Semi-DA. The bound suggests a principled way to obtain target generalization, i.e. by aligning both the marginal and conditional distributions across domains in feature space. Motivated by this, we then introduce the LIRR algorithm for jointly \textbf{L}earning \textbf{I}nvariant \textbf{R}epresentations and \textbf{R}isks. Finally, extensive experiments are conducted on both classification and regression tasks, which demonstrates LIRR consistently achieves state-of-the-art performance and significant improvements compared with the methods that only learn invariant representations or invariant risks.

Y. H. Tsai*, M. Q. Ma*, M. Yang,

In Proceedings of the 9th International Conference on Learning Representations (

[abs] [pdf]

This paper introduces Relative Predictive Coding (RPC), a new contrastive representation learning objective that maintains a good balance among training stability, minibatch size sensitivity, and downstream task performance. The key to the success of RPC is two-fold. First, RPC introduces the relative parameters to regularize the objective for boundedness and low variance. Second, RPC contains no logarithm and exponential score functions, which are the main cause of training instability in prior contrastive objectives. We empirically verify the effectiveness of RPC on benchmark vision and speech self-supervised learning tasks. Lastly, we relate RPC with mutual information (MI) estimation, showing RPC can be used to estimate MI with low variance.

P. Li, Y. Wang,

In Proceedings of the 9th International Conference on Learning Representations (

[abs] [pdf]

Disparate impact has raised serious concerns in machine learning applications and its societal impacts. In response to the need of mitigating discrimination, fairness has been regarded as a crucial property in algorithmic design. In this work, we study the problem of disparate impact on graph-structured data. Specifically, we focus on dyadic fairness, which articulates a fairness concept that a predictive relationship between two instances should be independent of the sensitive attributes. Based on this, we theoretically relate the graph connections to dyadic fairness on link predictive scores in learning graph neural networks, and reveal that regulating weights on existing edges in a graph contributes to dyadic fairness conditionally. Subsequently, we propose our algorithm, \textbf{FairAdj}, to empirically learn a fair adjacency matrix with proper graph structural constraints for fair link prediction, and in the meanwhile preserve predictive accuracy as much as possible. Empirical validation demonstrates that our method delivers effective dyadic fairness in terms of various statistics, and at the same time enjoys a favorable fairness-utility tradeoff.

J. Shen,

In Proceedings of the 34th Advances in Neural Information Processing Systems (

[abs] [pdf] [code] [video] [slides]

Model-based reinforcement learning methods learn a dynamics model with real data sampled from the environment and leverage it to generate simulated data to derive an agent. However, due to the potential distribution mismatch between simulated data and real data, this could lead to degraded performance. Despite much effort being devoted to reducing this distribution mismatch, existing methods fail to solve it explicitly. In this paper, we investigate how to bridge the gap between real and simulated data due to inaccurate model estimation for better policy optimization. To begin with, we first derive a lower bound of the expected return, which naturally inspires a bound maximization algorithm by aligning the simulated and real data distributions. To this end, we propose a novel model-based reinforcement learning framework AMPO, which introduces unsupervised model adaptation to minimize the integral probability metric (IPM) between feature distributions from real and simulated data. Instantiating our framework with Wasserstein-1 distance gives a practical model-based approach. Empirically, our approach achieves state-of-the-art performance in terms of sample efficiency on a range of continuous control benchmark tasks.

Y. H. Tsai,

In Proceedings of the 34th Advances in Neural Information Processing Systems (

[abs] [pdf] [video] [slides]

Since its inception, the neural estimation of mutual information (MI) has demonstrated the empirical success of modeling expected dependency between high-dimensional random variables. However, MI is an aggregate statistic and cannot be used to measure point-wise dependency between different events. In this work, instead of estimating the expected dependency, we focus on estimating point-wise dependency (PD), which quantitatively measures how likely two outcomes co-occur. We show that we can naturally obtain PD when we are optimizing MI neural variational bounds. However, optimizing these bounds is challenging due to its large variance in practice. To address this issue, we develop two methods (free of optimizing MI variational bounds): Probabilistic Classifier and Density-Ratio Fitting. We demonstrate the effectiveness of our approaches in 1) MI estimation, 2) self-supervised representation learning, and 3) cross-modal retrieval task.

R. Combes*,

In Proceedings of the 34th Advances in Neural Information Processing Systems (

[abs] [pdf] [code] [video] [slides]

Adversarial learning has demonstrated good performance in the unsupervised domain adaptation setting, by learning domain-invariant representations. However, recent work has shown limitations of this approach when label distributions differ between the source and target domains. In this paper, we propose a new assumption, \textit{generalized label shift} ($\glsa$), to improve robustness against mismatched label distributions. $\glsa$ states that, conditioned on the label, there exists a representation of the input that is invariant between the source and target domains. Under $\glsa$, we provide theoretical guarantees on the transfer performance of any classifier. We also devise necessary and sufficient conditions for $\glsa$ to hold, by using an estimation of the relative class weights between domains and an appropriate reweighting of samples. Our weight estimation method could be straightforwardly and generically applied in existing domain adaptation (DA) algorithms that learn domain-invariant representations, with small computational overhead. In particular, we modify three DA algorithms, JAN, DANN and CDAN and evaluate their performance on standard and artificial DA tasks. Our algorithms outperform the base versions, with vast improvements for large label distribution mismatches. Our code is available at \url{https://tinyurl.com/y585xt6j}.

In Proceedings of the 34th Advances in Neural Information Processing Systems (

[abs] [pdf] [video] [slides] [poster]

Crowdsourced data used in machine learning services might carry sensitive information about attributes that users do not want to share. Various methods have been proposed to minimize the potential information leakage of sensitive attributes while maximizing the task accuracy. However, little is known about the theory behind these methods. In light of this gap, we develop a novel theoretical framework for attribute obfuscation. Under our framework, we propose a minimax optimization formulation to protect the given attribute and analyze its inference guarantees against worst-case adversaries. Meanwhile, there is a tension between minimizing information leakage and maximizing task accuracy. To understand this, we prove an information-theoretic lower bound to precisely characterize the fundamental trade-off between accuracy and information leakage. We conduct experiments on two real-world datasets to corroborate the inference guarantees and validate the inherent trade-offs therein. Our results indicate that, among several alternatives, the adversarial learning approach achieves the best trade-off in terms of attribute obfuscation and accuracy maximization.

S. Zhao, X. Yue, S. Zhang, B. Li,

IEEE Transactions on Neural Networks and Learning Systems (

[abs] [pdf]

Large-scale labeled training datasets have enabled deep neural networks to excel across a wide range of benchmark vision tasks. However, in many applications, it is prohibitively expensive and time-consuming to obtain large quantities of labeled data. To cope with limited labeled training data, many have attempted to directly apply models trained on a large-scale labeled source domain to another sparsely labeled or unlabeled target domain. Unfortunately, direct transfer across domains often performs poorly due to the presence of domain shift or dataset bias. Domain adaptation is a machine learning paradigm that aims to learn a model from a source domain that can perform well on a different (but related) target domain. In this paper, we review the latest single-source deep unsupervised domain adaptation methods focused on visual tasks and discuss new perspectives for future research. We begin with the definitions of different domain adaptation strategies and the descriptions of existing benchmark datasets. We then summarize and compare different categories of single-source unsupervised domain adaptation methods, including discrepancy-based methods, adversarial discriminative methods, adversarial generative methods, and self-supervision-based methods. Finally, we discuss future research directions with challenges and possible solutions.

In Proceedings of the 37th International Conference on Machine Learning (

[abs] [pdf] [video] [slides] [blog]

The goal of universal machine translation is to learn to translate between any pair of languages, given a corpus of paired translated documents for a small subset of all pairs of languages. Despite impressive empirical results and an increasing interest in massively multilingual models, theoretical analysis on translation errors made by such universal machine translation models is only nascent. In this paper, we formally prove certain impossibilities of this endeavour in general, as well as prove positive results in the presence of additional (but natural) structure of data. For the former, we derive a lower bound on the translation error in the many-to-one translation setting, which shows that any algorithm aiming to learn shared sentence representations among multiple language pairs has to make a large translation error on at least one of the translation tasks, if no assumption on the structure of the languages is made. For the latter, we show that if the paired documents in the corpus follow a natural encoder-decoder generative process, we can expect a natural notion of ``generalization'': a linear number of language pairs, rather than quadratic, suffices to learn a good representation. Our theory also explains what kinds of connection graphs between pairs of languages are better suited: ones with longer paths result in worse sample complexity in terms of the total number of documents per language pair needed. We believe our theoretical insights and implications contribute to the future algorithmic design of universal machine translation.

P. Li,

In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (

[abs] [pdf]

Fair clustering aims to hide sensitive attributes during data partition by balancing the distribution of protected subgroups in each cluster. Existing work attempts to address this problem by reducing it to a classical balanced clustering with a constraint on the proportion of protected subgroups of the input space. However, the input space may limit the clustering performance, and so far only low-dimensional datasets have been considered. In light of these limitations, in this paper, we propose Deep Fair Clustering (DFC) to learn fair and clustering-favorable representations for clustering simultaneously. Our approach could effectively filter out sensitive attributes from representations, and also lead to representations that are amenable for the following cluster analysis. Theoretically, we show that our fairness constraint in DFC will not incur much loss in terms of several clustering metrics. Empirically, we provide extensive experimental demonstrations on four visual datasets to corroborate the superior performance of the proposed approach over existing fair clustering and deep clustering methods on both cluster validity and fairness criterion.

W. Wang,

In Proceedings of The Web Conference 2020 (

[abs] [pdf]

Early identification of patients at risk for postoperative complications can facilitate timely workups and treatments and improve health outcomes. Currently, a widely-used surgical risk calculator online web system developed by the American College of Surgeons (ACS) uses patients’ static features, e.g. gender, age, to assess the risk of postoperative complications. However, the most crucial signals that reflect the actual postoperative physical conditions of patients are usually real-time dynamic signals, including the vital signs of patients (e.g., heart rate, blood pressure) collected from postoperative monitoring. In this paper, we develop a dynamic postoperative complication risk scoring framework (DyCRS) to detect the “at-risk” patients in a real-time way based on postoperative sequential vital signs and static features. DyCRS is based on adaptations of the Hidden Markov Model (HMM) that captures hidden states as well as observable states to generate a real-time, probabilistic, complication risk score. Evaluating our model using electronic health record (EHR) on elective Colectomy surgery from a major health system, we show that DyCRS significantly outperforms the state-of-the-art ACS calculator and real-time predictors with 50.16% area under precision-recall curve (AUCPRC) gain on average in terms of detection effectiveness. In terms of earliness, our DyCRS can predict 15hrs55mins earlier on average than clinician’s diagnosis with the recall of 60% and precision of 55%. Furthermore, Our DyCRS can extract interpretable patients’ stages, which are consistent with previous medical postoperative complication studies. We believe that our contributions demonstrate significant promise for developing a more accurate, robust and interpretable postoperative complication risk scoring system, which can benefit more than 50 million annual surgeries in the US by substantially lowering adverse events and healthcare costs.

In Proceedings of the 8th International Conference on Learning Representations (

NeurIPS 2019 Workshop on Machine Learning with Guarantees (

[abs] [pdf] [video] [slides] [code]

We propose a novel algorithm for learning fair representations that can simultaneously mitigate two notions of disparity among different demographic subgroups. Two key components underpinning the design of our algorithm are balanced error rate and conditional alignment of representations. We show how these two components contribute to ensuring accuracy parity and equalized false-positive and false-negative rates across groups without impacting demographic parity. Furthermore, we also demonstrate both in theory and on two real-world experiments that the proposed algorithm leads to a better utility-fairness trade-off on balanced datasets compared with existing algorithms on learning fair representations.

T. Adel,

In Proceedings of the 8th International Conference on Learning Representations (

[abs] [pdf] [video]

Approaches to continual learning aim to successfully learn a set of related tasks that arrive in an online manner. Recently, several frameworks have been developed which enable deep learning to be deployed in this learning scenario. A key modelling decision is to what extent the architecture should be shared across tasks. On the one hand, separately modelling each task avoids catastrophic forgetting but it does not support transfer learning and leads to large models. On the other hand, rigidly specifying a shared component and a task-specific part enables task transfer and limits the model size, but it is vulnerable to catastrophic forgetting and restricts the form of task-transfer that can occur. Ideally, the network should adaptively identify which parts of the network to share in a data driven way. Here we introduce such an approach called Continual Learning with Adaptive Weights (CLAW), which is based on probabilistic modelling and variational inference. Experiments show that CLAW achieves state-of-the-art performance on six benchmarks in terms of overall continual learning performance, as measured by classification accuracy, and in terms of addressing catastrophic forgetting.

In Proceedings of the 33rd Advances in Neural Information Processing Systems (

[abs] [pdf] [poster] [slides] [blog]

With the prevalence of machine learning in high-stakes applications, especially the ones regulated by anti-discrimination laws or societal norms, it is crucial to ensure that the predictive models do not propagate any existing bias or discrimination. Due to the ability of deep neural nets to learn rich representations, recent advances in algorithmic fairness have focused on learning fair representations with adversarial techniques to reduce bias in data while preserving utility simultaneously. In this paper, through the lens of information theory, we provide the first result that quantitatively characterizes the tradeoff between demographic parity and the joint utility across different population groups. Specifically, when the base rates differ between groups, we show that any method aiming to learn fair representations admits an information-theoretic lower bound on the joint error across these groups. To complement our negative results, we also prove that if the optimal decision functions across different groups are close, then learning fair representations leads to an alternative notion of fairness, known as the accuracy parity, which states that the error rates are close between groups. Finally, our theoretical findings are also confirmed empirically on real-world datasets.

In NeurIPS 2019 Workshop on Machine Learning with Guarantees (

[abs] [pdf]

With the prevalence of machine learning services, crowdsourced data containing sensitive information poses substantial privacy challenges. Existing work focusing on protecting against membership inference attacks under the rigorous framework of differential privacy are vulnerable to attribute inference attacks. In light of the current gap between theory and practice, we develop a novel theoretical framework for privacy-preservation under the attack of attribute inference. Under our framework, we propose a minimax optimization formulation to protect the given attribute and analyze its privacy guarantees against arbitrary adversaries. On the other hand, it is clear that privacy constraint may cripple utility when the protected attribute is correlated with the target variable. To this end, we also prove an information-theoretic lower bound to precisely characterize the fundamental trade-off between utility and privacy. Empirically, we extensively conduct experiments to corroborate our privacy guarantee and validate the inherent trade-offs in different privacy preservation algorithms. Our experimental results indicate that the adversarial representation learning approaches achieve the best trade-off in terms of privacy preservation and utility maximization.

In Proceedings of the 36th International Conference on Machine Learning (

[abs] [pdf] [supplement] [poster] [slides] [blog]

Due to the ability of deep neural nets to learn rich representations, recent advances in unsupervised domain adaptation have focused on learning domain-invariant features that achieve a small error on the source domain. The hope is that the learnt representation, together with the hypothesis learnt from the source domain, can generalize to the target domain. In this paper, we first construct a simple counterexample showing that, contrary to common belief, the above conditions are not sufficient to guarantee successful domain adaptation. In particular, the counterexample (Fig. 1) exhibits \emph{conditional shift}: the class-conditional distributions of input features change between source and target domains. To give a sufficient condition for domain adaptation, we propose a natural and interpretable generalization upper bound that explicitly takes into account the aforementioned shift. Moreover, we shed new light on the problem by proving an information-theoretic lower bound on the joint error of \emph{any} domain adaptation method that attempts to learn invariant representations. Our result characterizes a fundamental tradeoff between learning invariant representations and achieving small joint error on both domains when the marginal label distributions differ from source to target. Finally, we conduct experiments on real-world datasets that corroborate our theoretical findings. We believe these insights are helpful in guiding the future design of domain adaptation and representation learning algorithms.

In Proceedings of the 33rd Advances in Neural Information Processing Systems (

[abs] [pdf] [poster] [slides] [code]

Feed-forward neural networks can be understood as a combination of an intermediate representation and a linear hypothesis. While most previous works aim to diversify the representations, we explore the complementary direction by performing an adaptive and data-dependent regularization motivated by the empirical Bayes method. Specifically, we propose to construct a matrix-variate normal prior (on weights) whose covariance matrix has a Kronecker product structure. This structure is designed to capture the correlations in neurons through backpropagation. Under the assumption of this Kronecker factorization, the prior encourages neurons to borrow statistical strength from one another. Hence, it leads to an adaptive and data-dependent regularization when training networks on small datasets. To optimize the model, we present an efficient block coordinate descent algorithm with analytical solutions. Empirically, we demonstrate that the proposed method helps networks converge to local optima with smaller stable ranks and spectral norms. These properties suggest better generalizations and we present empirical results to support this expectation. We also verify the effectiveness of the approach on multiclass classification and multitask regression problems with various network structures.

In Proceedings of the 35th Conference on Uncertainty in Artificial Intelligence (

Also In Learning with Limited Labeled Data: Weak Supervision and Beyond workshop at NIPS 2017

[abs] [pdf] [supplement] [poster]

We consider a multitask learning problem, in which several predictors are learned jointly. Prior research has shown that learning the relations between tasks, and between the input features, together with the predictor, can lead to better generalization and interpretability, which proved to be useful for applications in many domains. In this paper, we consider a formulation of multitask learning that learns the relationships both between tasks and between features, represented through a task covariance and a feature covariance matrix, respectively. First, we demonstrate that existing methods proposed for this problem present an issue that may lead to ill-posed optimization. We then propose an alternative formulation, as well as an efficient algorithm to optimize it. Using ideas from optimization and graph theory, we propose an efficient coordinate-wise minimization algorithm that has a closed form solution for each block subproblem. Our experiments show that the proposed optimization method is orders of magnitude faster than its competitors. We also provide a nonlinear extension that is able to achieve better generalization than existing methods.

Y. Xu*,

In Proceedings of the 28th International Joint Conference on Artificial Intelligence (

[abs] [pdf] [supplement] [Full arXiv version]

We consider peer review under a conference setting where there are conflicts between the reviewers and the submissions. Under such conflicts, reviewers can manipulate their reviews in a strategic manner to influence the final rankings of their own papers. Present-day peer-review systems are not designed to guard against such strategic behavior, beyond minimal (and insufficient) checks such as not assigning a paper to a conflicted reviewer. In this work, we address this problem through the lens of social choice, and present a theoretical framework for strategyproof and efficient peer review. Given the conflict graph which satisfies a simple property, we first present and analyze a flexible framework for reviewer-assignment and aggregation for the reviews that guarantees not only strategyproofness but also a natural efficiency property (unanimity). Our framework is based on the so-called partitioning method, and can be treated as a generalization of this type of method to conference peer review settings. We then empirically show that the requisite property on the (authorship) conflict graph is indeed satisfied in the ICLR-17 submissions data, and further demonstrate a simple trick to make the partitioning method more practically appealing under conference peer-review settings. Finally, we complement our positive results with negative theoretical results where we prove that under slightly stronger requirements, it is impossible for any algorithm to be both strategyproof and efficient.

C. Liang, J. Ye,

In Proceedings of the 12th International Conference on Educational Data Mining (

[abs] [pdf]

Strict partial order is a mathematical structure commonly seen in relational data. One obstacle to extracting such type of relations at scale is the lack of large scale labels for building effective data-driven solutions. We develop an active learning framework for mining such relations subject to a strict order. Our approach incorporates relational reasoning not only in finding new unlabeled pairs whose labels can be deduced from an existing label set, but also in devising new query strategies that consider the relational structure of labels. Our experiments on concept prerequisite relations show our proposed framework can substantially improve the classification performance with the same query budget compared to other baseline approaches.

In Proceedings of the 18th International Conference on Autonomous Agents and Multiagent Systems (

[abs] [pdf] [poster]

The ability to adapt to and learn from different domains and environments is crucial for agents to generalize. In this paper we propose a probabilistic framework for domain adaptation that blends both generative and discriminative modeling in a principled way. Under this framework, generative and discriminative models correspond to specific choices of the prior over parameters. By maximizing both the marginal and the conditional log-likelihoods, our models can use both labeled instances from the source domain as well as unlabeled instances from \emph{both} source and target domains. We show that the popular reconstruction loss of autoencoder corresponds to an upper bound of the negative marginal log-likelihoods of unlabeled instances, and give a generalization bound that explicitly incorporates it into the analysis. We instantiate our framework using neural networks, and build a concrete model, DAuto.

In Proceedings of the 32nd Advances in Neural Information Processing Systems (

[abs] [pdf] [supplement] [poster] [code]

While domain adaptation has been actively researched, most algorithms focus on the single-source-single-target adaptation setting. In this paper we propose new generalization bounds and algorithms under both classification and regression settings for unsupervised multiple source domain adaptation. Our theoretical analysis naturally leads to an efficient learning strategy using adversarial neural networks: we show how to interpret it as learning feature representations that are invariant to the multiple domain shifts while still being discriminative for the learning task. To this end, we propose multisource domain adversarial networks (MDAN) that approach domain adaptation by optimizing task-adaptive generalization bounds. To demonstrate the effectiveness of MDAN, we conduct extensive experiments showing superior adaptation performance on both classification and regression problems: sentiment analysis, digit classification, and vehicle counting.

In Proceedings of the 34th Conference on Uncertainty in Artificial Intelligence (

[abs] [pdf] [supplement]

Symmetric nonnegative matrix factorization has found abundant applications in various domains by providing a symmetric low-rank decomposition of nonnegative matrices. In this paper we propose a Frank-Wolfe (FW) solver to optimize the symmetric nonnegative matrix factorization problem under a simplicial constraint, which has recently been proposed for probabilistic clustering. Compared with existing solutions, this algorithm is simple to implement, and has no hyperparameters to be tuned. Building on the recent advances of FW algorithms in nonconvex optimization, we prove an $O(1/\eps^2)$ convergence rate to $\eps$-approximate KKT points, via a tight bound $\Theta(n^2)$ on the curvature constant, which matches the best known result in unconstrained nonconvex setting using gradient methods. Numerical results demonstrate the effectiveness of our algorithm. As a side contribution, we construct a simple nonsmooth convex problem where the FW algorithm fails to converge to the optimum. This result raises an interesting question about necessary conditions of the success of the FW algorithm on convex problems.

IEEE International Conference on Acoustics, Speech and Signal Processing (

[abs] [pdf] [slides]

We propose an end-to-end model based on convolutional and recurrent neural networks for speech enhancement. Our model is purely data-driven and does not make any assumptions about the type or the stationarity of the noise. In contrast to existing methods that use multilayer perceptrons (MLPs), we employ both convolutional and recurrent neural network architectures. Thus, our approach allows us to exploit local structures in both the frequency and temporal domains. By incorporating prior knowledge of speech signals into the design of model structures, we build a model that is more data-efficient and achieves better generalization on both seen and unseen noise. Based on experiments with synthetic data, we demonstrate that our model outperforms existing methods, improving PESQ by up to 0.6 on seen noise and 0.64 on unseen noise.

In Uncertainty in Deep Learning workshop at UAI (

[abs] [pdf] [poster]

We propose an approximate empirical Bayes framework and an efficient algorithm for learning the weight matrix of deep neural networks. Empirically, we show the proposed method works as a regularization approach that helps generalization when training neural networks on small datasets.

In 6th International Conference on Learning Representations (

[abs] [pdf] [poster]

While domain adaptation has been actively researched in recent years, most theoretical results and algorithms focus on the single-source-single-target adaptation setting. Naive application of such algorithms on multiple source domain adaptation problem may lead to suboptimal solutions. We propose a new generalization bound for domain adaptation when there are multiple source domains with labeled instances and one target domain with unlabeled instances. Compared with existing bounds, the new bound does not require expert knowledge about the target distribution, nor the optimal combination rule for multisource domains. Interestingly, our theory also leads to an efficient learning strategy using adversarial neural networks: we show how to interpret it as learning feature representations that are invariant to the multiple domain shifts while still being discriminative for the learning task. To this end, we propose two models, both of which we call multisource domain adversarial networks (MDANs): the first model optimizes directly our bound, while the second model is a smoothed approximation of the first one, leading to a more data-efficient and task-adaptive model. The optimization tasks of both models are minimax saddle point problems that can be optimized by adversarial training. To demonstrate the effectiveness of MDANs, we conduct extensive experiments showing superior adaptation performance on three real-world datasets: sentiment analysis, digit classification, and vehicle counting.

In Proceedings of the 31st Advances in Neural Information Processing Systems (

[abs] [pdf] [poster]

Bayesian online algorithms for Sum-Product Networks (SPNs) need to update their posterior distribution after seeing one single additional instance. To do so, they must compute moments of the model parameters under this distribution. The best existing method for computing such moments scales quadratically in the size of the SPN, although it scales linearly for trees. This unfortunate scaling makes Bayesian online algorithms prohibitively expensive, except for small or tree-structured SPNs. We propose a linear-time algorithm that works even when the SPN is a general directed acyclic graph (DAG). Our algorithm significantly broadens the applicability of Bayesian online algorithms for SPNs. There are three key ingredients in the design and analysis of our algorithm: 1). For each edge in the graph, we find a linear time reduction from the moment computation problem to a joint inference problem in SPNs. 2). Using the property that each SPN computes a multilinear polynomial, we construct an efficient procedure for polynomial evaluation by differentiation without expanding the network that may contain exponentially many positive monomials. 3). We propose a dynamic programming method to further reduce the computation of the moments of all the edges in the graph from quadratic to linear. We demonstrate the usefulness of our linear time moment computation algorithm by applying it to develop a linear time assume density filter (ADF) for SPNs.

T. Adel,

In Proceedings of the 31th AAAI Conference on Artificial Intelligence (

[abs] [pdf]

Domain adaptation addresses learning tasks where training is performed on data from one domain whereas testing is performed on data belonging to a different but related domain. Assumptions about the relationship between the source and target domains should lead to tractable solutions on the one hand, and be realistic on the other hand. Here we propose a generative domain adaptation model that allows for modeling different assumptions about this relationship, among which is a newly introduced assumption that replaces covariate shift with a possibly more realistic assumption without losing tractability due to the efficient variational inference procedure developed. In addition to the ability to model less restrictive relationships between source and target, modeling can be performed without any target labeled data (unsupervised domain adaptation). We also provide a Rademacher complexity bound of the proposed algorithm. We evaluate the model on the Amazon reviews and the CVC pedestrian detection datasets.

Y. H. Tsai,

In Time Series workshop at NIPS (

[abs] [pdf] [slides] [poster]

The assumption that data samples are independently identically distributed is the backbone of many learning algorithms. Nevertheless, datasets often exhibit rich structures in practice, and we argue that there exist some unknown orders within the data instances. Aiming to find such orders, we introduce a novel Generative Markov Network (GMN) which we use to extract the order of data instances automatically. Specifically, we assume that the instances are sampled from a Markov chain. Our goal is to learn the transitional operator of the chain as well as the generation order by maximizing the generation probability under all possible data permutations. One of our key ideas is to use neural networks as a soft lookup table for approximating the possibly huge, but discrete transition matrix. This strategy allows us to amortize the space complexity with a single model and make the transitional operator generalizable to unseen instances. To ensure the learned Markov chain is ergodic, we propose a greedy batch-wise permutation scheme that allows fast training. Empirically, we evaluate the learned Markov chain by showing that GMNs are able to discover orders among data instances and also perform comparably well to state-of-the-art methods on the one-shot recognition benchmark task.

In Proceedings of the 30th Advances in Neural Information Processing Systems (

[abs] [pdf] [supplement] [poster] [code]

We present a unified approach for learning the parameters of Sum-Product networks (SPNs). We prove that any complete and decomposable SPN is equivalent to a mixture of trees where each tree corresponds to a product of univariate distributions. Based on the mixture model perspective, we characterize the objective function when learning SPNs based on the maximum likelihood estimation (MLE) principle and show that the optimization problem can be formulated as a signomial program. Both the projected gradient descent (PGD) and the exponentiated gradient (EG) in this setting can be viewed as first order approximations of the signomial program after proper transformation of the objective function. Based on the signomial program formulation, we construct two parameter learning algorithms for SPNs by using sequential monomial approximations (SMA) and the concave-convex procedure (CCCP), respectively. The two proposed methods naturally admit multiplicative updates, hence effectively avoiding the projection operation. With the help of the unified framework, we also show that, in the case of SPNs, CCCP leads to the same algorithm as Expectation Maximization (EM) despite the fact that they are different in general. Extensive experiments on 20 data sets demonstrate the effectiveness and efficiency of the two proposed approaches for learning SPNs. We also show that the proposed methods can improve the performance of structure learning and yield state-of-the-art results.

In Proceedings of the 33rd International Conference on Machine Learning (

[abs] [pdf] [poster] [slides] [code]

Sum-Product Networks (SPNs) are probabilistic inference machines that admit exact inference in linear time in the size of the network. Existing parameter learning approaches for SPNs are largely based on the maximum likelihood principle and hence are subject to overfitting compared to more Bayesian approaches. Exact Bayesian posterior inference for SPNs is computationally intractable. Both standard variational inference and posterior sampling for SPNs are computationally infeasible even for networks of moderate size due to the large number of local latent variables per instance. In this work, we propose a novel deterministic collapsed variational inference algorithm for SPNs that is computationally efficient, easy to implement and at the same time allows us to incorporate prior information into the optimization formulation. Extensive experiments show a significant improvement in accuracy compared with a maximum likelihood based approach.

P. Jaini, A. Rashwan,

In Proceedings of the 8th International Conference on Probabilistic Graphical Models (

[abs] [pdf]

Sum-product networks (SPNs) have recently emerged as an attractive representation due to their dual interpretation as a special type of deep neural network with clear semantics and a tractable probabilistic graphical model. We explore online algorithms for parameter learning in SPNs with continuous variables. More specifically, we consider SPNs with Gaussian leaf distributions and show how to derive an online Bayesian moment matching algorithm to learn from streaming data. We compare the resulting generative models to stacked restricted Boltzmann machines and generative moment matching networks on real-world datasets.

A. Rashwan,

In Proceedings of the 19th International Conference on Artificial Intelligence and Statistics (

[abs] [pdf]

Probabilistic graphical models provide a general and flexible framework for reasoning about complex dependencies in noisy domains with many variables. Among the various types of probabilistic graphical models, sum-product networks (SPNs) have recently generated some interest because exact inference can always be done in linear time with respect to the size of the network. This is particularly attractive since it means that learning an SPN from data always yields a tractable model for inference. However, existing parameter learning algorithms for SPNs operate in batch mode and do not scale easily to large datasets. In this work, we explore online algorithms to ensure that parameter learning can also be done tractably with respect to the amount of data. More specifically, we propose a new Bayesian moment matching (BMM) algorithm that operates naturally in an online fashion and that can be easily distributed. We demonstrate the effectiveness and scalability of BMM in comparison to online extensions of gradient descent and expectation maximization on 20 classic benchmarks and 4 large scale datasets.

In Proceedings of the 32nd International Conference on Machine Learning (

[abs] [pdf] [supplement] [Full arXiv version] [slides] [poster]

In this paper, we establish some theoretical connections between Sum-Product Networks (SPNs) and Bayesian Networks (BNs). We prove that every SPN can be converted into a BN in linear time and space in terms of the network size. The key insight is to use Algebraic Decision Diagrams (ADDs) to compactly represent the local conditional probability distributions at each node in the resulting BN by exploiting context-specific independence (CSI). The generated BN has a simple directed bipartite graphical structure. We show that by applying the Variable Elimination algorithm (VE) to the generated BN with ADD representations, we can recover the original SPN where the SPN can be viewed as a history record or caching of the VE inference process. To help state the proof clearly, we introduce the notion of {\em normal} SPN and present a theoretical analysis of the consistency and decomposability properties. We conclude the paper with some discussion of the implications of the proof and establish a connection between the depth of an SPN and a lower bound of the tree-width of its corresponding BN.

In Proceedings of the 24th International Joint Conference on Artificial Intelligence (

[abs] [pdf] [slides] [poster] [code]

The ability to accurately model a sentence at varying stages (e.g., word-phrase-sentence) plays a central role in natural language processing. As an effort towards this goal we propose a self-adaptive hierarchical sentence model (AdaSent). AdaSent effectively forms a hierarchy of representations from words to phrases and then to sentences through recursive gated local composition of adjacent segments. We design a competitive mechanism (through gating networks) to allow the representations of the same sentence to be engaged in a particular learning task (e.g., classification), therefore effectively mitigating the gradient vanishing problem persistent in other recursive models. Both qualitative and quantitative analysis shows that AdaSent can automatically form and select the representations suitable for the task at hand during training, yielding superior classification performance over competitor models on 5 benchmark data sets.

In Proceedings of the 29th AAAI Conference on Artificial Intelligence (

[abs] [pdf] [poster]

We propose SoF (Soft-cluster matrix Factorization), a probabilistic clustering algorithm which softly assigns each data point into clusters. Unlike model-based clustering algorithms, SoF does not make assumptions about the data density distribution. Instead, we take an axiomatic approach to define 4 properties that the probability of co-clustered pairs of points should satisfy. Based on the properties, SoF utilizes a distance measure between pairs of points to induce the conditional co-cluster probabilities. The objective function in our framework establishes an important connection between probabilistic clustering and constrained symmetric Nonnegative Matrix Factorization (NMF), hence providing a theoretical interpretation for NMF-based clustering algorithms. To optimize the objective, we derive a sequential minimization algorithm using a penalty method. Experimental results on both synthetic and real-world datasets show that SoF significantly outperforms previous NMF-based algorithms and that it is able to detect non-convex patterns as well as cluster boundaries.

F. Faisal,

IEEE/ACM Transactions on Computational Biology and Bioinformatics (

In Proceedings of the 4th ACM International Conference on Bioinformatics, Computational Biology and Biomedicine (

[abs] [pdf] [supplement]

Analogous to sequence alignment, network alignment (NA) can be used to transfer biological knowledge across species between conserved network regions. NA faces two algorithmic challenges: 1) Which cost function to use to capture “similarities” between nodes in different networks? 2) Which alignment strategy to use to rapidly identify “high-scoring” alignments from all possible alignments? We “break down” existing state-of-the-art methods that use both different cost functions and different alignment strategies to evaluate each combination of their cost functions and alignment strategies. We find that a combination of the cost function of one method and the alignment strategy of another method beats the existing methods. Hence, we propose this combination as a novel superior NA method. Then, since human aging is hard to study experimentally due to long lifespan, we use NA to transfer aging-related knowledge from well annotated model species to poorly annotated human. By doing so, we produce novel human aging-related knowledge, which complements currently available knowledge about aging that has been obtained mainly by sequence alignment. We demonstrate significant similarity between topological and functional properties of our novel predictions and those of known aging-related genes. We are the first to use NA to learn more about aging.

In Method of Moments and Spectral Learning workshop at ICML (

[abs] [pdf] [slides] [poster] [code]

Spectral learning recently generated lots of excitement in machine learning, largely because it is the first known method to produce consistent estimates (under suitable conditions) for several latent variable models. In contrast, maximum likelihood estimates may get trapped in local optima due to the non-convex nature of the likelihood function of latent variable models. In this paper, we do an empirical evaluation of spectral learning (SL) and expectation maximization (EM), which reveals an important gap between the theory and the practice. First, SL often leads to negative probabilities. Second, EM often yields better estimates than spectral learning and it does not seem to get stuck in local optima. We discuss how the rank of the model parameters and the amount of training data can yield negative probabilities. We also question the common belief that maximum likelihood estimators are necessarily inconsistent.

Fundamental Limits and Tradeoffs in Invariant Representation Learning
H. Zhao*, C. Dan*, B. Aragam, T. Jaakkola, G. Gordon, and P. Ravikumar
arXiv preprint [abs] [pdf] Many machine learning applications involve learning representations that achieve two competing goals: To maximize information or accuracy with respect to a subset of features (e.g.\ for prediction) while simultaneously maximizing invariance or independence with respect to another, potentially overlapping, subset of features (e.g.\ for fairness, privacy, etc). Typical examples include privacy-preserving learning, domain adaptation, and algorithmic fairness, just to name a few. In fact, all of the above problems admit a common minimax game-theoretic formulation, whose equilibrium represents a fundamental tradeoff between accuracy and invariance. Despite its abundant applications in the aforementioned domains, theoretical understanding on the limits and tradeoffs of invariant representations is severely lacking. In this paper, we provide an information-theoretic analysis of this general and important problem under both classification and regression settings. In both cases, we analyze the inherent tradeoffs between accuracy and invariance by providing a geometric characterization of the feasible region in the information plane, where we connect the geometric properties of this feasible region to the fundamental limitations of the tradeoff problem. In the regression setting, we also derive a tight lower bound on the Lagrangian objective that quantifies the tradeoff between accuracy and invariance. This lower bound leads to a better understanding of the tradeoff via the spectral properties of the joint distribution. In both cases, our results shed new light on this fundamental problem by providing insights on the interplay between accuracy and invariance. These results deepen our understanding of this fundamental problem and may be useful in guiding the design of adversarial representation learning algorithms. |

I enjoy sketching and calligraphy at my spare time. If I have a long vacation, I also enjoy traveling. |