In order to participate in the seminar, a few minutes prior to the beginning of the seminar you should connect to the Zoom conference via the following link or manually using the Zoom conference ID 897 7646 2466 and password 549526. If you have problems logging into Zoom, you can connect to the YouTube livestream of the seminarMachine learning provides an alternative to the classical physics model building approach. Interatomic interaction models (also known as interatomic potentials or force fields) are an example of an area where machine learning has proven to be indispensable. In my talk, I will tell you how the application of machine learning ideas allows you to build models in accuracy comparable to quat-mechanical models, but superior in performance by several orders of magnitude. I will also show how the ideas of active learning (also known as "optimal experimental design") can automate the process of constructing interatomic potentials, which paves the way for the creation of autonomous algorithms that automate some actions, for example, the analysis of convergence of computational methods, which traditionally should be performed by users of algorithms.
