Neural Networks for Mathematical Discovery

Our world is full of surprises, and the same thing can be said of the world of mathematics. Surprises like the icosahedron, aperiodic tilings and the Mandelbrot set have captivated the imaginations of mathematicians and led to new mathematical fields.

In this Presidential Lecture, Geordie Williamson will survey recent work using neural networks to aid the search for interesting examples and counterexamples. He says this work is just the tip of the iceberg, predicting that the coming years will see a flourishing of AI-assisted experimental mathematics.