SOURCE-LINKED INTELLIGENCE

Steering Equilibrium Selection in Regularized Self-Play via the Reference Policy

arXiv · Artificial Intelligence · article · Sep 17, 2026 · UTC

Regularized self-play -- the family behind DeepNash's Stratego play -- drives a two-player zero-sum policy to a Nash equilibrium by best-responding to a slowly moving, entropy-regularized reference policy $ρ$. When the game has a polytope of value-equivalent equilibria, the regularizer silently breaks the tie: with a uniform reference it selects the maximum-entropy member, the I-projection of $ρ$ onto the Nash set. Can the reference be used to choose the equilibrium on purpose? On five exactly solvable games plus a 2-D polytope, with exact best responses and equivalence tests over independent

Read original source ↗ Open in workspace

recordType

paper

region

Global