Dear Friends,
I am pleased to announce that our SLDS July webinar will be held on July 23, 2 pm Eastern Time, featured by Dr. Alex Luedtke from Harvard. Hope to see you there!
Title: DoubleGen: Debiased Generative Modeling of Counterfactuals
Speakers: Dr. Alex Luedtke, Harvard University
Date and Time: July 23, 2026, 2:00 to 3:30 pm Eastern Time
Registration Link: ASA SLDS Webinar Registration Link [eventbrite.com]
Abstract:
Generative models for counterfactual outcomes face two key sources of bias. Confounding bias arises when approaches fail to account for systematic differences between those who receive the intervention and those who do not. Misspecification bias arises when methods attempt to address confounding through estimation of an auxiliary model, but specify it incorrectly. We introduce DoubleGen, a doubly robust framework that modifies generative modeling training objectives to mitigate these biases. The new objectives rely on two auxiliaries -- a propensity and outcome model -- and successfully address confounding bias even if only one of them is correct. We provide finite-sample guarantees for this robustness property. We further establish conditions under which DoubleGen achieves oracle optimality -- matching the convergence rates standard approaches would enjoy if interventional data were available -- and minimax rate optimality. We illustrate DoubleGen with three examples: diffusion models, flow matching, and autoregressive language models.
Presenter:
Alex Luedtke is a Professor of Health Care Policy and Affiliate in Statistics at Harvard University. Previously, he held faculty appointments in Statistics and Biostatistics at the University of Washington and Fred Hutch.
Alex works at the intersection of semiparametrics, causal inference, and machine learning. Much of his work leverages computational power—be it through automatic differentiation, deep learning, or game theory—to automate the construction of efficient statistical procedures. He applies these methods as part of multidisciplinary teams to improve decision-making in mental health and infectious disease research.
Selected honors include a COPSS Emerging Leader Award, NIH Director's New Innovator Award, and AWS Machine Learning Research Award. His work has been supported by NIH, NSF, PCORI, WHO, Netflix, and Amazon.
As a mentor, Alex has supervised the completed dissertations of 8 PhD students, several of whom now hold tenure-track positions in statistics and biostatistics.
Alex received a PhD in Biostatistics from UC Berkeley and an ScB in Applied Mathematics from Brown University.
------------------------------
Boxiang Wang
Associate Professor
Department of Statistics and Actuarial Science
University of Iowa
Iowa City, IA, United States
------------------------------