Applying Machine Learning to Number Theory

In the late 18th century, Gauss conjectured the prime number theorem by computing large tables of prime numbers. In 1965, Birch and Swinnerton arrived at the BSD conjecture via computer-aided computations. Contemporary mathematicians have access to large-scale digital computation and modern machine-learning techniques, allowing computers to aid both computation and pattern recognition. We will describe how these tools can aid mathematics research, demonstrate how they led to a recent discovery in number theory, and outline the proof of a theorem that emerged from these machine-learning experiments. No number theory background is required.