Machine Learning for the Materials World

Efthimios (Tim) Kaxiras
Harvard University

The last few years have witnessed a surge of activity in machine learning approaches applied to materials science, boosted in part by President Obama’s Materials Genome Initiative. While there is great promise, there are also pitfalls in applying data science and machine learning methods to the discovery of new materials. In this talk I will address both the promise and the pitfalls on using data science ideas to explore the possibilities of “materials by design”, drawing on examples from our recent research. Applications of our work focus on new materials for energy related problems, including improved batteries, photovoltaics, and new catalysts; on a different level, we have also explored fundamental questions like the strength of amorphous solids.

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