Home /  Workshop /  Schedules /  Distribution Generalization in Underidentified Causal Models

Distribution Generalization in Underidentified Causal Models

[Virtual] Hot Topics: Foundations of Stable, Generalizable and Transferable Statistical Learning March 07, 2022 - March 10, 2022

March 08, 2022 (09:00 AM PST - 09:25 AM PST)
Speaker(s): Jonas Peters (University of Copenhagen)
Location: SLMath: Online/Virtual
Tags/Keywords
  • causality

  • distribution generalization

  • robustness

  • identifiability

  • intervention

Primary Mathematics Subject Classification
Secondary Mathematics Subject Classification No Secondary AMS MSC
Video

Distribution Generalization In Underidentified Causal Models

Abstract

We consider the problem of predicting a response Y from a set of covariates X when test and training distributions differ. We consider a setting where such differences have causal explanations and the test distributions emerge from interventions. Causal models minimize the worst-case risk under arbitrary interventions on the covariates but may not always be identifiable from observational or interventional data. In this talk, we argue that underidentification and distribution generalization are closely connected. We propose to consider most predictive invariant models and discuss some of their properties. We also present limits of distribution generalization.

Supplements
92772?type=thumb Distribution Generalization in Underidentified Causal Models 1.71 MB application/pdf Download
Video/Audio Files

Distribution Generalization In Underidentified Causal Models

Troubles with video?

Please report video problems to itsupport@slmath.org.

See more of our Streaming videos on our main VMath Videos page.