6 December 2021

Jakob Macke, University of Tubingen, Germany
Maximilian Dax, Max Planck Institute for Intelligent Systems, Germany

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 simulation based inference methods using flexible density estimators parameterised with neural networks, our efforts on benchmarking these approaches, and applications to modelling problems in neuroscience and beyond. In addition, Maximilian Dax (a Phd student in the group of Bernhard Schölkopf, MPI for Intelligent Systems Tübingen) will show how simulation-based inference can be used to achieve unprecedented performance on a challenging application from astrophysics, and makes it possible to infer the parameters of gravitational wave models in a few seconds, without any drop in accuracy relative to MCMC methods which take multiple hours for inference.

In cooperation with the Jean Golding Institute, University of Bristol

More information: the Bristol Data Science Seminar Series