Glossary · Term

anchored gradient descent-ascent

← all terms

Definition

A known optimization algorithm for saddle-point problems whose convergence rate an AI agent tightened with a new parameter schedule.

A first-order method for convex-concave saddle-point optimization; in the AI Co-Mathematician work, a Lean-proof-search agent discovered a new parameter schedule that provably improves its convergence rate.

Mentioned in 1 episode

  1. 067
    An AI Just Solved a 1996 Erdős Problem—and the Simplest Agent Won