The Computational and Statistical Material Science Lab | Lead: Kristofer Reyes

About the lab


Our lab works at the interface of computation, mathematics, statistics, and materials science. We are specifically interested in meaningful machine learning in the small-data regime that dominates much of materials science. In this regime, we must be:

  1. Economical with the limited data that we do have;
  2. Strategic in acquiring new data; and
  3. Resourceful by incorporating knowledge and information from other sources.

We work on Bayesian models, decision-making under uncertainty, reinforcement learning, and methods to fuse data with physics-based knowledge and expert opinion. We do not view materials science as a black-box source of “yet-another-dataset”. Instead, the field offers a rich problem setting filled with nuance and structure. We develop methods that appreciate this nuance and utilize such structure.

More information about the lab


People

Research

Publications

Teaching

Project pages


Holmes: a framework for decision-making under uncertainty

Kris’s Discount Wheels: key references for ML/AI for materials

What’s new


CSMS Lab (Twitter: @CSMSLab | Email: [email protected]) | Department of Materials Design and Innovation | 134 Bell Hall, University at Buffalo

Scratch