Jordan Ellenberg: What does machine learning have to offer number theory? (NTWS 221)

Abstract: This is going to be a somewhat informal where I report on some of my own work, some work of others, and some stuff I’m learning about at the ongoing Harvard/CMSA workshop on machine learning in mathematics. I will focus on an outlook where the goal is not to reproduce or replace our central enterprise of writing proofs of theorems and understanding things, but rather on models for machine-human collaboration, where ML techniques are used to generate interesting hypotheses, examples, and ideas as a kind of force multiplier for traditional mathematicians. I’ll probably talk about cap sets, computing GCDs, murmurations, navigating Cayley graphs, and probably some other stuff besides! (Note: Oct 7-11 will be a number theory week at CMSA so any questions the audience wants to suggest we work on there are very welcome!)
Link to slides: drive.google.c...
Number Theory Web Seminar: www.ntwebsemin...
Original air date:
Thursday, September 26, 2024 (8am PDT, 11am EDT, 4pm BST, 5pm CEST, 6pm Israel Daylight Time, 8:30pm Indian Standard Time, 11pm CST)
Friday, September 27, 2024 (1am AEST, 3am NZST)