Machine Learning for Materials Design: Combination of Theoretical methods, Heuristics, and Hybrid Techniques

Sadasivan Shankar
Harvard University

The challenges facing Materials Design in the long route from material discovery to real
applications include synthesis techniques, measurement capabilities, theoretical methods, and advanced computational methods. In this talk, we will illustrate application of a combination of methods including theoretical methods, heuristics, and hybrid methods for model reduction and practical application of materials design. The talk will also touch upon key aspects of machine learning including selection of descriptors, development of methods to use these descriptors for model reduction, and specific applications with practical examples. The applications will span areas of complex chemical processing, measurement techniques, and properties between interfaces of materials. In addition, we will try to map the role of machine learning in 4 different aspects in Materials Design. In addition, we will also talk about prediction methods and properties covering quantum, mescoscopic, and continuum scales.

Back to Workshop IV: Synergies between Machine Learning and Physical Models