Machine Learning for Knot Theory

Traditionally, machine learning — due to its ability to detect structure in large datasets — has largely been a tool for applied mathematics.  Recent work has repurposed these ML tools to reveal new relationships in pure mathematics, aiding human intuition and even guiding conjecture generation. We discuss how knot theory presents itself as a playground for these ML methods, and experiment with invariant prediction, classification tasks, and even applications to biopolymer identification.

For more information, visit the Math Table website.