Speaker: Prof. Dr. Niki Kilbertus

Date: 26.11.2024, 16:00-17:00

Location: SAP Labs Munich (MUE03), Friedrich-Ludwig-Bauer-Straße 5, 85748 Garching bei München, Auditorium (AE.76)

Abstract: In this talk, we will work out the differences between causal relationships and mere associational relationships. We will see how to formalize this difference and provide first glimpses into how we can sometimes estimate causal relationships from purely observational data in the instrumental variable framework. Finally, we take these ideas further to develop an iterative method for experiment design.
This talk will require at least a basic understanding of ML, but could then be quite interesting, because most of the time we would actually like to understand causal dependencies rather than mere correlations to make useful decisions. The beginning would be rather general, whereas in the latter parts I would be showcasing some of our recent research papers.

Bio: Niki Kilbertus holds a joint appointment as assistant professor at the Technical University of Munich and as group leader at Helmholtz AI. His research interests revolve broadly around causality and machine learning systems for robust mechanistic insights. Niki obtained a PhD in Machine Learning as a Cambridge-Tübingen fellow in a joint program of the University of Cambridge and the Max Planck Institute for Intelligent Systems. During his PhD, he interned at Google, Deepmind, and Amazon. Originally, his background is in mathematics and physics in which he holds Master’s degrees from the University of Regensburg. During his studies, he also spent time at Harvard and Stanford as a visiting researcher.