Materials Informatics
t-SNE embedding of materials data
Project Features
- Consulting with scientists and engineers across industry and universities
- Using Grama to investigate materials data
- Deploying novel data science techniques to accelerate materials discovery
Project Description
Materials informatics is the use of data science to support the study of materials. One of the most exciting topics in materials informatics is the use of machine learning to discover novel high-performing materials: The usual paradigm of machine learning is to accurately predict previously-seen cases from representative data. With materials discovery we are actively seeking outliers—new materials that perform better than anything previously seen. To make progress in this challenging space scientists have to combine data science with quantified uncertainties, and find ways to leverage domain knowledge and physical intuition.
I use my background in statistics and data science to develop novel model-assessment techniques, and work as a consultant to help discover novel materials faster.
Students working on this project will need some experience with coding in Python. Project members will use machine learning and data science techniques to support materials research.
Related Works
Zachary del Rosario
Assistant Professor of Engineering and Applied Statistics
Empowering scientists and engineers to reason under uncertainty
