Statistical Rule Learning for Materials Science

Explore a seminar on statistical rule learning for materials science presented by Mario Boley, a senior lecturer in Data Science and AI at Monash University. Delve into the challenges of applying machine learning to materials science, including issues with model opacity and poor extrapolation. Learn how rule-based models can improve the field by detecting domains of applicability for complex models and directly modeling material properties. Discover how these interpretable machine learning approaches can drive hypothesis formation and guide targeted data acquisition in materials research. Gain insights into Boley's background in computer science and his work on interpretable modeling and data-driven knowledge discovery, particularly in the context of novel materials discovery. Understand the limitations of current machine learning approaches in materials science and the potential for rule-based models to provide a more transparent and physically-informed layer to materials modeling.