The Vendi Score: A Diversity Metric For Science And Machine Learning

Measuring diversity correctly is important for science and machine learning. In this talk, I'll describe the Vendi Score, a diversity metric that connects and extends ideas from ecology and quantum mechanics. The Vendi Score is applicable in any domain where similarity can be defined. I'll showcase the usefulness of the Vendi Score in efficiently exploring molecular conformation spaces and in enabling the discovery of novel MOFs and organic photoredox catalysts.