Abstract: Many fields of science make extensive use of mechanistic forward models which are implemented through numerical simulators, requiring the use of simulation-based approaches to statistical inference. I will talk about our recent work on developing and benchmarking simulation based inference methods using flexible density estimators parameterised with neural networks. In particular, I will talk about our practical experiences in using simulation-based inference approaches in applications in neuroscience, computational imaging and astrophysics.

Speakers: Jakob Macke
Jakob has been Professor for “Machine Learning in Science” at the University of Tübingen, Germany, since May 2020. The W3 professorship has been set up as part of the Cluster of Excellence “Machine Learning: New Perspectives for the Sciences”. He is also an Adjunct Research Scientist at the Max Planck Institute for Intelligent Systems, Director of the Bernstein Center for Computational Neuroscience, and an ELLIS Fellow and member of the ELLIS Unit Tübingen.

Affiliation: Tübingen University

Place of Seminar:  Zoom