ANC Seminar - Stefan Bauer

Speaker: Stefan Bauer of Max Planck Institute for Intelligent Systems

Title: Towards Causal Representation Learning

Abstract:

The two fields of machine learning and graphical causality arose and developed separately. Still, there is strong cross-pollination now, and increasing interest in both fields to benefit from the advances of the other. We review fundamental concepts of causal inference and relate them to crucial open problems of machine learning, including transfer and generalization, arguing that causality can contribute to modern machine learning research. This also holds in the opposite direction: we note that most work in causality starts from the premise that the causal variables are observed. A central problem for AI and causality is, thus, causal representation learning, the discovery of high-level causal variables from low-level observations. Finally, we delineate some implications of causality for machine learning and propose key research areas and benchmarks at the intersection of both communities.

Bio: Stefan Bauer is a research group leader at the MPI for Intelligent Systems and a CIFAR Azrieli Global Scholar. He has a BSc. and MSc. in Mathematics as well as a PhD in Computer Science from ETH Zurich. During his studies he held scholarships from the Swiss and German National Merit Foundation and his PhD was awarded with an ETH Medal for an outstanding doctoral thesis. In 2019 he won the best paper award at the International Conference of Machine Learning (ICML).