Portrait of Patrik Reizinger, Ascent Fellow 2027
2027 ASCENT FELLOW

Patrik Reizinger

IVADO postdoctoral fellow
Université de Montréal and Mila
Trustworthy AI (Interpretability, Theory)

Highlighted research

Identifiable Exchangeable Mechanisms for Causal Structure and Representation Learning

ICLR 2025, Spotlight

Biography

Patrik Reizinger is an IVADO postdoctoral fellow at the Université de Montréal and Mila, co-supervised by Dhanya Sridhar and Simon Lacoste-Julien. He did his Ph.D. research in the Robust Machine Learning group at the Max Planck Institute for Intelligent Systems in Tübingen, supervised by Wieland Brendel, Ferenc Huszár, Matthias Bethge, and Bernhard Schölkopf. He has also spent time at the Vector Institute and at the University of Cambridge. The main motivation for his research is to advance our understanding of how and why deep learning models work. His research studies identifiable causal representation learning and out-of-distribution generalization, with a focus on compositionality and interpretability in language models. He has been building tools to improve and better understand AI-assisted research, including a research skills marketplace for AI agents, a hallucination detection benchmark and a detection tool. He is a fellow of the Big-if-True Science accelerator, run by Renaissance Philanthropy with SPRIND, where he designs a research programme for automatic, graded verification of scientific claims.