LGGTMAOCJun 6, 2022

Policy Optimization for Markov Games: Unified Framework and Faster Convergence

Salesforce
arXiv:2206.02640v431 citationsh-index: 73
Originality Incremental advance
AI Analysis

This work addresses the problem of slow convergence in multi-agent reinforcement learning algorithms for researchers and practitioners, offering incremental improvements over existing methods.

This paper tackles policy optimization for multi-agent reinforcement learning in Markov Games by proposing a unified algorithmic framework that converges to approximate Nash equilibria, achieving faster convergence rates of up to O~(T^{-1}) compared to the previous best O~(T^{-1/2}) for two-player zero-sum games and extending to multi-player general-sum games with O~(T^{-3/4}) rates.

This paper studies policy optimization algorithms for multi-agent reinforcement learning. We begin by proposing an algorithm framework for two-player zero-sum Markov Games in the full-information setting, where each iteration consists of a policy update step at each state using a certain matrix game algorithm, and a value update step with a certain learning rate. This framework unifies many existing and new policy optimization algorithms. We show that the state-wise average policy of this algorithm converges to an approximate Nash equilibrium (NE) of the game, as long as the matrix game algorithms achieve low weighted regret at each state, with respect to weights determined by the speed of the value updates. Next, we show that this framework instantiated with the Optimistic Follow-The-Regularized-Leader (OFTRL) algorithm at each state (and smooth value updates) can find an $\mathcal{\widetilde{O}}(T^{-5/6})$ approximate NE in $T$ iterations, and a similar algorithm with slightly modified value update rule achieves a faster $\mathcal{\widetilde{O}}(T^{-1})$ convergence rate. These improve over the current best $\mathcal{\widetilde{O}}(T^{-1/2})$ rate of symmetric policy optimization type algorithms. We also extend this algorithm to multi-player general-sum Markov Games and show an $\mathcal{\widetilde{O}}(T^{-3/4})$ convergence rate to Coarse Correlated Equilibria (CCE). Finally, we provide a numerical example to verify our theory and investigate the importance of smooth value updates, and find that using "eager" value updates instead (equivalent to the independent natural policy gradient algorithm) may significantly slow down the convergence, even on a simple game with $H=2$ layers.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes