Angela Zhou

401S Bridge Hall

I am an Assistant Professor at USC Marshall Data Sciences and Operations, in the Operations group.

Previously I was a research fellow at the Simons program on causality, a FODSI postdoc at UC Berkeley, hosted by Bin Yu, and Michael I. Jordan. I obtained my PhD from Cornell University in Operations Research and Information Engineering working with Nathan Kallus at Cornell Tech. My work was previously supported on a NDSEG fellowship.

My research interests are broadly in data-driven decision making under uncertainty, including operations, statistical machine learning, and causal inference, and the interplay of statistics and optimization.

I’m happy to chat about possible collaborations. If you are a student at USC, please feel free to email me (and check with your advisor). If you are interested in becoming a PhD student at USC, please apply to the Data Sciences and Operations PhD program if you’d like to work with me.

Email: zhoua at usc.edu.

[Scholar], [CV],

Research:

Computerized influence functions for optimization-based estimators in causal inference
Credible causal inference for optimal decisions and bounds; and for infinite-horizon offline reinforcement learning under violations of assumptions
Algorithmic fairness: biased data, partial identification bounds on causal disparities. Substantive: don’t use CJ data for ML benchmarks, evaluating short term impacts of bail reform on aggregate crime.

selected publications

Empirical Gateaux Derivatives for Causal Inference

Michael Jordan, Yixin Wang, and Angela Zhou

conference version at Neurips; journal version in preparation 2022

Abs arXiv

We study a constructive algorithm that approximates Gateaux derivatives for statistical functionals by finite-differencing, with a focus on causal inference functionals. We consider the case where probability distributions are not known a priori but also need to be estimated from data. These estimated distributions lead to empirical Gateaux derivatives, and we study the relationships between empirical, numerical, and analytical Gateaux derivatives. Starting with a case study of counterfactual mean estimation, we instantiate the exact relationship between finite-differences and the analytical Gateaux derivative. We then derive requirements on the rates of numerical approximation in perturbation and smoothing that preserve the statistical benefits of one-step adjustments, such as rate-double-robustness. We then study more complicated functionals such as dynamic treatment regimes and the linear-programming formulation for policy optimization in infinite-horizon Markov decision processes. The newfound ability to approximate bias adjustments in the presence of arbitrary constraints illustrates the usefulness of constructive approaches for Gateaux derivatives. We also find that the statistical structure of the functional (rate-double robustness) can permit less conservative rates of finite-difference approximation. This property, however, can be specific to particular functionals, e.g. it occurs for the counterfactual mean but not the infinite-horizon MDP policy value.
Minimax-Optimal Policy Learning under Unobserved Confounding

Nathan Kallus, and Angela Zhou

Management Science (Forthcoming), supersedes Neurips 2018 version 2020

Abs Code

We study the problem of learning personalized decision policies from observational data while accounting for possible unobserved confounding in the data-generating process. Unlike previous approaches that assume unconfoundedness, ie, no unobserved confounders affected both treatment assignment and outcomes, we calibrate policy learning for realistic violations of this unverifiable assumption with uncertainty sets motivated by sensitivity analysis in causal inference. Our framework for confounding-robust policy improvement optimizes the minimax regret of a candidate policy against a baseline or reference" status quo" policy, over an uncertainty set around nominal propensity weights. We prove that if the uncertainty set is well-specified, robust policy learning can do no worse than the baseline, and only improve if the data supports it. We characterize the adversarial subproblem and use efficient algorithmic solutions to optimize over parametrized spaces of decision policies such as logistic treatment assignment. We assess our methods on synthetic data and a large clinical trial, demonstrating that confounded selection can hinder policy learning and lead to unwarranted harm, while our robust approach guarantees safety and focuses on well-evidenced improvement.
Confounding-Robust Policy Evaluation in Infinite-Horizon Reinforcement Learning

Nathan Kallus, and Angela Zhou

Neurips 2020

Abs arXiv Code

Off-policy evaluation of sequential decision policies from observational data is necessary in applications of batch reinforcement learning such as education and healthcare. In such settings, however, observed actions are often confounded with transitions by unobserved variables, rendering exact evaluation of new policies impossible, ie, unidentifiable. We develop a robust approach that estimates sharp bounds on the (unidentifiable) value of a given policy in an infinite-horizon problem given data from another policy with unobserved confounding subject to a sensitivity model. We phrase the problem precisely as computing the support function of the set of all stationary state-occupancy ratios that agree with both the data and the sensitivity model. We show how to express this set using a new partially identified estimating equation and prove convergence to the sharp bounds, as we collect more confounded data. We prove that membership in the set can be checked by solving a linear program, while the support function is given by a difficult nonconvex optimization problem. We leverage an analytical solution for the finite-state-space case to develop approximations based on nonconvex projected gradient descent. We demonstrate the resulting bounds empirically.
Assessing algorithmic fairness with unobserved protected class using data combination

Nathan Kallus, Xiaojie Mao, and Angela Zhou

Management Science (Forthcoming). A preliminary version appeared at FaCCT 2020 2019

Abs arXiv Code

The increasing impact of algorithmic decisions on people’s lives compels us to scrutinize their fairness and, in particular, the disparate impacts that ostensibly-color-blind algorithms can have on different groups. Examples include credit decisioning, hiring, advertising, criminal justice, personalized medicine, and targeted policymaking, where in some cases legislative or regulatory frameworks for fairness exist and define specific protected classes. In this paper we study a fundamental challenge to assessing disparate impacts in practice: protected class membership is often not observed in the data. This is particularly a problem in lending and healthcare. We consider the use of an auxiliary dataset, such as the US census, that includes class labels but not decisions or outcomes. We show that a variety of common disparity measures are generally unidentifiable aside for some unrealistic cases, providing a new perspective on the documented biases of popular proxy-based methods. We provide exact characterizations of the sharpest-possible partial identification set of disparities either under no assumptions or when we incorporate mild smoothness constraints. We further provide optimization-based algorithms for computing and visualizing these sets, which enables reliable and robust assessments–an important tool when disparity assessment can have far-reaching policy implications. We demonstrate this in two case studies with real data: mortgage lending and personalized medicine dosing.

news

Oct 7, 2022	Co-chaired ACM EAAMO; catch up with the program here
Sep 14, 2022	Two papers accepted at Neurips 2022: Off-Policy Evaluation with Policy-Dependent Response and Empirical Gateaux Derivatives for Causal Inference (Oral-designated, Top 3% of All Submissions).
Jun 16, 2022	Started at USC Marshall.