Home / Research by Area

Research by Area

Papers organized by research theme. For a chronological listing, see Methodological Work and Review, Workshop Publication, and Applied Work on the home page.

Causal Discovery

Cross-theme connections

  • Regularization Paths for Continuous DAG Learning with Diyang Li, Fei Wang (NeurIPS 2026)

  • MixCIT: A Kernel Based Local-Polynomial Debiased Test for Conditional Independence on Mixed-Type Data with Mengxiao Gao, Promit Ghosal (arXiv)
    CI / hypothesis testing; primary use case is causal discovery.

  • From Guess2Graph: When and How Can Unreliable Experts Safely Boost Causal Discovery in Finite Samples? with Sujai Hiremath, Dominik Janzing, Philipp Faller, Patrick Blobaum, Elke Kirschbaum, Shiva Prasad Kasiviswanathan (arXiv) (AISTATS 2026)

  • When Additive Noise Meets Unobserved Mediators: Bivariate Denoising Diffusion for Causal Discovery with Dominik Meier, Sujai Hiremath, Promit Ghosal (arXiv) (NeurIPS 2025)

  • LoSAM: Local Search in Additive Noise Models with Mixed Mechanisms and General Noise for Global Causal Discovery with Sujai Hiremath, Promit Ghosal (arXiv) (UAI 2025)

  • Local Causal Discovery for Structural Evidence of Direct Discrimination with Jacqueline Maasch, Violet Chen, Agni Orfanoudaki, Nil-Jana Akpinar, Fei Wang (arXiv) (AAAI 2025)
    LD3 returns a set optimal for weighted controlled (direct) effect. Related: Optimal Adjustment Sets

  • Hybrid Top-Down Global Causal Discovery with Local Search for Linear and Nonlinear Additive Noise Models with Sujai Hiremath, Jaqueline Maasch, Mengxiao Gao, Promit Ghosal (arXiv) (NeurIPS 2024)

  • Local Discovery by Partitioning: Polynomial-Time Causal Discovery Around Exposure-Outcome Pairs with Jacqueline Maasch, Weishen Pan, Shantanu Gupta, Volodymyr Kuleshov, Fei Wang (arXiv) (UAI 2024) (NeurIPS Causal Representation Learning Workshop 2023)
    LDP — local discovery aimed at causal inference (with LD3)

Causal, Longitudinal & Sequential Inference

Cross-theme connections

Longitudinal

  • Deep Doubly Debiased Longitudinal Effect Estimation with ICE G-Computation with Wenxin Chen, Weishen Pan, Fei Wang (arXiv) (KDD 2026)

  • Smooth Multi-Policy Causal Effect Estimation in Longitudinal Settings with Wenxin Chen, Weishen Pan, Fei Wang (arXiv) (ICML 2026)

Static

  • Federated Targeted Maximum Likelihood Estimation with Diyang Li, Fei Wang (arXiv)

  • Never Go Full Batch: Stochastic TMLE for Large-Scale Debiased Inference with Diyang Li, Fei Wang (NeurIPS 2026)

  • MOSIC: Model-Agnostic Optimal Subgroup Identification with Multi-Constraint for Improved Reliability with Wenxin Chen, Weishen Pan, Fei Wang (arXiv) (ICML 2026)

  • Targeted Maximum Likelihood Learning: An Optimization Perspective with Diyang Li (NeurIPS 2025) (a copy here)

  • Optimal Adjustment Sets for Nonparametric Estimation of Weighted Controlled Direct Effect with Ruiyang Lin, Yongyi Guo (arXiv) (NeurIPS 2025)
    LD3 returns a set optimal for this WCDE estimand. Related: LD3

  • Kernel Debiased Plug-in Estimation: Simultaneous, Automated Debiasing without Influence Functions for Many Target Parameters with Brian Cho, Yaroslav Mukhin, and Ivana Malenica (arXiv) (ICML 2024)
    Finalist, 2023 INFORMS DMDA Workshop Best Paper Competition -- Theoretical Track

  • Causal Inference with Selectively Deconfounded Data with Andrew Li, Zachary Lipton, and Sridhar Tayur (AISTATS 2021) (Journal version) (NeurIPS CausalML Workshop 2019)

  • Federated Causal Inference for Effect Estimation in Healthcare: Methods, Challenges, and Applications with Haoyang Li, Zheqi Lyu, Meihan Wei, Simai Cheng, Wehan Zheng, Fuying Ding, Jie Xu, Max Sunong, Colin Magdamo, Mark Albers, Fei Wang, Chengxi Zang (arXiv) (npj Artificial Intelligence 2026)

Sequential

  • Reward Maximization for Pure Exploration: Minimax Optimal Good Arm Identification for Nonparametric Multi-Armed Bandits with Brian Cho, Dominik Meier, Nathan Kallus (arXiv) (AISTATS 2025)

  • CSPI-MT: Calibrated Safe Policy Improvement with Multiple Testing for Threshold Policies with Brian Cho, Ana-Roxana Pop, Sam Corbett-Davies, Israel Nir, Ariel Evnine, Nathan Kallus (arXiv) (KDD 2025)

  • Peeking with PEAK: Sequential, Nonparametric Composite Hypothesis Tests for Means of Multiple Data Streams with Brian Cho and Nathan Kallus (arXiv) (ICML 2024)

  • Anytime-Valid Inference in N-of-1 Trials with Ivana Malenica, Yongyi Guo, and Stefan Konigorski (arXiv) (ML4H 2023)

Adaptive Experimentation and Reinforcement Learning

Cross-theme connections

  • Integrating Causal DAGs in Deep RL: Activating Minimal Markovian States with Multi-Order Exposure with Jiamin Xu, Jacqueline Maasch (arXiv)
    Bridges causal structure / discovery with deep RL.

  • Fast Non-Episodic Finite-Horizon RL with K-Step Lookahead Thresholding with Jiamin Xu (arXiv) (ICML 2026)

  • From Restless to Contextual: A Thresholding Bandit Approach to Improve Finite-Horizon Performance with Jiamin Xu, Ivan Nazarov, Aditya Rastogi, Africa Perianez (arXiv) (AISTATS 2026)

  • Online Uniform Sampling: Randomized Learning-Augmented Approximation Algorithms with Application to Digital Health with Xueqing Liu, Esmaeil Keyvanshokooh, and Susan A. Murphy (arXiv)

  • Learning When to Nudge: A Dual-Agent Bandit Framework for Behavioral Interventions with Esmaeil Keyvanshokooh, Yongyi Guo, Xueqing Liu, and Susan A. Murphy (Journal version)

  • Contextual Bandits with Budgeted Information Reveal with Esmaeil Keyvanshokooh, Xueqing Liu, and Susan A. Murphy (arXiv) (AISTATS 2024)

  • Toward a Liquid Biopsy: Greedy Approximation Algorithms for Active Sequential Hypothesis Testing with Su Jia, Andrew Li, and Sridhar Tayur (SSRN) (Management Science, 2025)

  • Greedy Approximation Algorithms for Active Sequential Hypothesis Testing with Su Jia, and Andrew Li (NeurIPS 2021)
    Winner, 2021 INFORMS Pierskalla Best Paper Award

  • Optimizing Wearable Devices in Personalized Opioid Use Disorder Treatments Under Budget Constraint with Yanhan (Savannah) Tang, Alan Scheller-Wolf and Sridhar Tayur (Journal version)
    Canadian Healthcare Optimization Workshop (CHOW) best paper in the category of operations research/management science, 2021
    Finalist, 2019 INFORMS IBM Service Science Best Student Paper Award

Generative Models

  • Cross-Modal Latent Influence: Patch-Level Training Sample Attribution for Vision-Language Models with Laura Yao, Shixing Yu, Jie Peng, Promit Ghosal, Tianlong Chen (arXiv)

  • Correcting Influence: Unboxing LLM Outputs with Orthogonal Latent Spaces with Shixing Yu, Promit Ghosal (arXiv)

  • Clustering by Denoising: Latent Plug-and-Play Diffusion for Single-Cell Data with Dominik Meier, Shixing Yu, Sagnik Nandy, Promit Ghosal (arXiv) (ICLR 2026)

Applications