KL regularization enables Õ(1/n) convergence for offline Nash equilibria in zero-sum Markov games under unilateral concentrability via the ROSE framework and SOS-MD algorithm.
A Deeper Look at Planning as Learning from Replay
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
years
2026 5verdicts
UNVERDICTED 5roles
background 2polarities
background 2representative citing papers
Establishes almost sure convergence rates arbitrarily close to o(n^{1-2η}) for power-law rates η in (1/2,1) and o(n^{-1}) for harmonic rates in contractive stochastic approximation with Markovian noise.
AstroAlertBench evaluates multimodal LLMs on astronomical classification accuracy, reasoning, and honesty using real ZTF alerts, revealing that high accuracy often diverges from self-assessed reasoning quality.
Proposes OPMD algorithm achieving accelerated O(1/n) rates for offline Nash equilibrium learning in alpha-potential games via reference-anchored data coverage.
KL regularization enables pessimism-free offline learning in general-sum games, recovering regularized Nash equilibria at accelerated rate O(1/n) via GANE and converging to coarse correlated equilibria at standard rate O(1/sqrt(n)+1/T) via GAMD.
citing papers explorer
-
Offline Two-Player Zero-Sum Markov Games with KL Regularization
KL regularization enables Õ(1/n) convergence for offline Nash equilibria in zero-sum Markov games under unilateral concentrability via the ROSE framework and SOS-MD algorithm.
-
Almost Sure Convergence Rates of Stochastic Approximation and Reinforcement Learning via a Poisson-Moreau Drift
Establishes almost sure convergence rates arbitrarily close to o(n^{1-2η}) for power-law rates η in (1/2,1) and o(n^{-1}) for harmonic rates in contractive stochastic approximation with Markovian noise.