Abstract:

Computational materials science is undergoing a second revolution empowered by machine learning (ML). ML methods do not completely reply on the theoretical understanding of the problem but take a data-driven approach to solve these problems. ML methods can describe and predict the notorious properties of materials, especially for those that can only be determined experimentally.

In this talk, I will present our works in applying ML to identify the degradation patterns of Li-ion batteries (Nat. Comms 11 (1), 1-6 (2020)) and design new high-temperature superconductors. I will show that combining advanced experimental technique and physical theory with ML can help us to understand the physical laws between materials features and properties from a new perspective.