UCPH Statistics Seminar: Julius von Kügelgen

Title: Backtracking Counterfactuals and Explainable AI

Speaker: Julius von Kügelgen from ETH Zürich

Abstract: Counterfactual reasoning---envisioning hypothetical scenarios, or possible worlds, where some circumstances are different from what (f)actually occurred (counter-to-fact)---is ubiquitous in human cognition. Conventionally, counterfactually altered circumstances have been treated as "small miracles" that locally violate the laws of nature while sharing the same initial conditions. In Pearl's structural causal model (SCM) framework, this is made mathematically rigorous via interventions that modify the causal laws while the values of exogenous variables are shared. In recent years, however, this purely interventionist account of counterfactuals has increasingly come under scrutiny from both philosophers and psychologists. Instead, they suggest a backtracking account of counterfactuals, according to which the causal laws remain unchanged in the counterfactual world; differences to the factual world are instead "backtracked" to altered initial conditions (exogenous variables). In this talk, we will explore and formalise this alternative mode of counterfactual reasoning within the SCM framework, discuss the backtracking semantics in the context of related literature, and draw connections to recent developments in explainable artificial intelligence (XAI) [1]. Finally, I will showcase an application of backtracking for XAI in the context of explanations for SCMs with deep generative model components [2]. 

References:
[1] https://arxiv.org/abs/2211.00472
[2] https://arxiv.org/abs/2310.07665

Bio: Julius (https://www.juliusvonkugelgen.com/) is a postdoc with Jonas Peters at the Seminar for Statistics at ETH Zürich. His research interests lie at the intersection of causal inference and machine learning. He did his PhD at the University of Cambridge and the Max Planck Institute for Intelligent Systems, Tübingen, advised by Bernhard Schölkopf and Adrian Weller. Previously, he studied Mathematics (B.Sc., M.Sci.) at Imperial College London (UK) and Artificial Intelligence (M.Sc.) at UPC Barcelona (Spain) and TU Delft (Netherlands).