Skip to main content

Machine learning properties of solid-state materials: Considerations for interpretability, reliability, and data-efficiency.

Date:
-
Location:
Zoom
Speaker(s) / Presenter(s):
Dr. Christopher Sutton

Abstract: Advances in machine learning (ML) are making a large impact in many disciplines, including materials and computational chemistry. A particularly exciting application of ML is the prediction of quantum mechanical (QM) properties (e.g., formation energy, bandgap, etc.) using only the structure as input. Assuming sufficient accuracies in the ML models, these methods enable screening of a considerably large chemical space at orders of magnitude lower computational cost than available QM methods. Despite the promise of ML in chemistry, several key challenges remain in both applying and interpreting the results of ML algorithms. Here, we will discuss our efforts in addressing these issues, including our recent work on opening the black box of ML methods by identifying the domain of applicability, i.e., where a given model is reliable.

Bio: Chris Sutton is an Assistant Professor in the Department of Chemistry & Biochemistry at the University of South Carolina. Chris received his PhD at the Georgia Institute of Technology under the direction of Professor Jean-Luc Bredas, and then moved to Duke University for postdoctoral research with Professor Weitao Yang. Chris received the Alexander von Humboldt postdoctoral fellowship to work in the Theory Department at the Fritz Haber Institute in Berlin, Germany where Matthias Scheffler was the Director. Chris’ current research is focused on computational materials discovery through a combination of   electronic structure calculations, machine learning, and stochastic sampling techniques to speed up the traditional computational design of materials.