AI Paper Atlas

Research map / Learning theory and algorithms / Submodular and game algorithms

Near-Optimal Reinforcement Learning with Self-Play

Yu Bai, Chi Jin, Tiancheng Yu

NeurIPS 2020 · Learning theory and algorithms › Submodular and game algorithms · approach: Zero-sum game learning dynamics · 13 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use optimistic Nash Q-learning instead of VI-ULCBSelf-play for approximate Nash equilibrium · Zero-sum Markov gamessample complexity (episodes): $\tlO(H^4 S^2 AB/\epsilon^2)$ -> $\tlO(H^5SAB/\epsilon^2)$Improved
Use optimistic Nash V-learning instead of optimistic Nash Q-learningSelf-play for approximate Nash equilibrium · Zero-sum Markov gamessample complexity (episodes): $\tlO(H^5SAB/\epsilon^2)$ -> $\quad\tlO(H^6S(A+B)/\epsilon^2)\quad$Mixed
Use Nash V-learning rather than Nash Q-learning when the action sets are sufficiently largeSelf-play for approximate Nash equilibrium · Zero-sum Markov games with $\min\{A,B\}>H$sample complexity (episodes): $\tlO(SAB)$ -> $\tlO(S(A+B))$Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy