Pith. sign in

REVIEW 4 cited by

When Can We Learn General-Sum Markov Games with a Large Number of Players Sample-Efficiently?

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2110.04184 v2 pith:DCJXHN4F submitted 2021-10-08 cs.LG cs.GTstat.ML

classification cs.LGcs.GTstat.ML
keywords epsilonlearninggamesequilibriumplayersalgorithmepisodesgeneral-sum
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

Multi-agent reinforcement learning has made substantial empirical progresses in solving games with a large number of players. However, theoretically, the best known sample complexity for finding a Nash equilibrium in general-sum games scales exponentially in the number of players due to the size of the joint action space, and there is a matching exponential lower bound. This paper investigates what learning goals admit better sample complexities in the setting of $m$-player general-sum Markov games with $H$ steps, $S$ states, and $A_i$ actions per player. First, we design algorithms for learning an $\epsilon$-Coarse Correlated Equilibrium (CCE) in $\widetilde{\mathcal{O}}(H^5S\max_{i\le m} A_i / \epsilon^2)$ episodes, and an $\epsilon$-Correlated Equilibrium (CE) in $\widetilde{\mathcal{O}}(H^6S\max_{i\le m} A_i^2 / \epsilon^2)$ episodes. This is the first line of results for learning CCE and CE with sample complexities polynomial in $\max_{i\le m} A_i$. Our algorithm for learning CE integrates an adversarial bandit subroutine which minimizes a weighted swap regret, along with several novel designs in the outer loop. Second, we consider the important special case of Markov Potential Games, and design an algorithm that learns an $\epsilon$-approximate Nash equilibrium within $\widetilde{\mathcal{O}}(S\sum_{i\le m} A_i / \epsilon^3)$ episodes (when only highlighting the dependence on $S$, $A_i$, and $\epsilon$), which only depends linearly in $\sum_{i\le m} A_i$ and significantly improves over existing efficient algorithm in the $\epsilon$ dependence. Overall, our results shed light on what equilibria or structural assumptions on the game may enable sample-efficient learning with many players.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Solving Zero-Sum Convex Markov Games

    cs.GT 2025-06 conditional novelty 7.0 of 10

    Independent policy-gradient algorithms provably compute approximate Nash equilibria in two-player zero-sum convex Markov games.

  2. Provable Partially Observable Reinforcement Learning with Privileged Information

    cs.LG 2024-12 conditional novelty 7.0 of 10

    The paper gives the first provable polynomial-sample and quasi-polynomial-time guarantees for expert distillation and belief-weighted asymmetric actor-critic in POMDPs with privileged state information.

  3. Incentivize without Bonus: Provably Efficient Model-based Online Multi-agent RL for Markov Games

    cs.LG 2025-02 conditional novelty 6.0 of 10

    Value-incentivized exploration via best-response values gives near-optimal regret for NE/CCE in linear-model Markov games without explicit uncertainty bonuses.

  4. Minimax-Optimal Multi-Agent Robust Reinforcement Learning

    cs.LG 2024-12 conditional novelty 6.0 of 10

    Robust Q-FTRL achieves ε-robust CCE in R-contaminated Markov games with H^3 S Σ_i A_i min{H,1/R}/ε^2 samples up to logs, matching a new lower bound; two-player zero-sum gives NE.

Pith tools