Machine Learning for Materials Discovery
Alex van Grootel, Data Scientist
Machine Learning (ML) has been applied to many manufacturing areas such as predictive maintenance or forecasting. However, ML for R&D has not seen as much attention. Materials researchers often work on problems with relatively high dimensions, and these problems are complicated by the fact that there are often multiple property constraints that have to be satisfied simultaneously. As a result, classical methods like Design of Experiments (DOE) or human intuition are inadequate. If instead we rely on more sophisticated tools like ML to drive progress, we observe significantly shorter innovation cycles which translates to much faster time to market. Using an approach that combines a graphical modeling framework with sequential learning and robust uncertainty quantification, Citrine has helped materials researchers dramatically accelerate the development new materials in domains as diverse as alloys, formulations and semi-conductors. In this talk, Alex van Grootel will discuss how we addressed challenges, and point out some remaining outstanding questions in the field.
Alex van Grootel is a Data Scientist at Citrine Informatics where he helps materials and chemicals companies to develop new materials using AI. Previously, Alex was a graduate researcher at MIT in the Olivetti Group where he was applying deep learning techniques to materials problems. Education: BEng University of Edinburgh (Mechanical Engineering), MEng MIT (Advanced Manufacturing & Design), SM MIT (Technology & Policy), SM MIT (Computer Science).