Mathematics & Machine Learning Seminar

Knot theorists have a long history of compiling tables of knots and their invariants. These invariants come from a variety of sources (including physics and biology) and involve constructions which are algebraic, analytic, geometric, and combinatorial in nature. Despite the subject's long history, there are many fundamental open questions about these invariants that have yet to be answered by classical techniques.

In recent years, machine learning has emerged as a viable approach to studying many of these problems. This approach raises new questions about the structures underlying these knot data sets. In this talk I will outline questions and results at the intersection of knot theory and ML, and describe approaches to using reinforcement learning to solve difficult generative problems in knot theory.