Machine Learning in Materials Science: Recent Progress and Critical Next Steps

Rampi Ramprasad
University of Connecticut

Data-to-knowledge ideas are beginning to show promise within materials science [1]. Among these ideas is “machine learning”, a branch of artificial intelligence pertaining to the creation of models that can effectively learn from past data, and make rapid predictions and decisions when confronted with new situations. In this seminar, I will provide a general background to this topic, and discuss a few examples where such methods may be used within materials science, such as discovering simple models describing complex materials behavior [2], screening for superior or new materials through accelerated property predictions [3], and acceleration of molecular dynamics simulations [4]. I will end with the critical challenges that the community is presently grappling with.
[1] T. Mueller, A. G. Kusne, R. Ramprasad, “Machine Learning in Materials Science: Recent Progress and Emerging Applications”, Reviews in Computational Chemistry (2016).
[2] C. Kim, G. Pilania, R. Ramprasad, “From Organized High-throughput Data to Phenomenological Theory using Machine Learning: The Example of Dielectric Breakdown”, Chem. Mater. 28, 1304 (2016).
[3] A. Mannodi-Kanakkithodi, G. Pilania, T. D. Huan, T. Lookman, R. Ramprasad, “Machine Learning Strategy for Accelerated Design of Polymer Dielectrics”, Sci. Rep., 6, 20952 (2016).
[4] V. Botu, R. Ramprasad, “Learning scheme to predict atomic forces and accelerate materials simulations”, Phys. Rev. B 92, 094306 (2015).

Presentation (PDF File)

Back to Workshop I: Machine Learning Meets Many-Particle Problems