University of Copenhagen
Title: Causality and Robust Prediction
Purely predictive methods have do not perform well when the test distribution changes too much from the training distribution. Causal models are known to be stable with respect to distributional shifts such as arbitrarily strong interventions on the covariates, but do not perform well when the test distribution differs only mildly from the training distribution. We introduce Anchor Regression, a framework that provides a tradeoff between causal and predictive models. The method poses different (convex and non-convex) optimization problems and relates to methods that are tailored for instrumental variable settings.
If time allows, we show how similar principles can be used for inferring metabolic networks. No prior knowledge about causality is required.
Jonas is a professor in statistics at the Department of Mathematical Sciences at the University of Copenhagen. Previously, he has been at the MPI for Intelligent Systems in Tubingen and was a Marie Curie fellow at the Seminar for Statistics, ETH Zurich. He studied Mathematics at the University of Heidelberg and the University of Cambridge. In his research, Jonas is interested in inferring causal relationships from different types of data and in building statistical methods that are robust with respect to distributional shifts. He seeks to combine theory, methodology, and applications.