paper-with-me

홈 › Papers

Assigning Credit with Partial Reward Decoupling in Multi-Agent Proximal Policy Optimization

2024-08-08 · Aditya Kapoor, Benjamin Freed, Howie Choset, Jeff Schneider

Multi-agent proximal policy optimization (MAPPO) has recently demonstrated state-of-the-art performance on challenging multi-agent reinforcement learning tasks. However, MAPPO still struggles with the credit assignment problem, wherein the sheer difficulty in ascribing credit to individual agents' actions scales poorly with team size. In this paper, we propose a multi-agent reinforcement learning algorithm that adapts recent developments in credit assignment to improve upon MAPPO. Our approach leverages partial reward decoupling (PRD), which uses a learned attention mechanism to estimate which of a particular agent's teammates are relevant to its learning updates. We use this estimate to dynamically decompose large groups of agents into smaller, more manageable subgroups. We empirically demonstrate that our approach, PRD-MAPPO, decouples agents from teammates that do not influence their expected future reward, thereby streamlining credit assignment. We additionally show that PRD-MAPPO yields significantly higher data efficiency and asymptotic performance compared to both MAPPO and other state-of-the-art methods across several multi-agent tasks, including StarCraft II. Finally, we propose a version of PRD-MAPPO that is applicable to \textit{shared} reward settings, where PRD was previously not applicable, and empirically show that this also leads to performance improvements over MAPPO.

📄 PDF Abstract BibTeX arXiv:2408.04295

Code (1)

uoe-agents/pressureplate 공식 구현

Tasks

Multi-agent Reinforcement Learningreinforcement-learningReinforcement LearningStarcraftStarcraft II

Methods 이 논문이 사용한 방법론

Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…
Attention 설명 없음

Similar Papers 제목 키워드 기반

Learning Cooperative Multi-Agent Policies with Partial Reward Decoupling

2021-12-23 · Benjamin Freed, Aditya Kapoor, Ian Abraham, Jeff Schneider 외

One of the preeminent obstacles to scaling multi-agent reinforcement learning to large numbers of agents is assigning credit to individual agents' actions. In this paper, we address this credit assignment problem with an…

counterfactualMulti-agent Reinforcement Learning

Beyond BLEU:Training Neural Machine Translation with Semantic Similarity

2019-07-01 · ACL 2019 7 · John Wieting, Taylor Berg-Kirkpatrick, Kevin Gimpel, Graham Neubig

While most neural machine translation (NMT)systems are still trained using maximum likelihood estimation, recent work has demonstrated that optimizing systems to directly improve evaluation metrics such as BLEU can signi…

DiversityMachine TranslationNMTSemantic Similarity+2

Beyond BLEU: Training Neural Machine Translation with Semantic Similarity

2019-09-14 · John Wieting, Taylor Berg-Kirkpatrick, Kevin Gimpel, Graham Neubig

While most neural machine translation (NMT) systems are still trained using maximum likelihood estimation, recent work has demonstrated that optimizing systems to directly improve evaluation metrics such as BLEU can subs…

DiversityMachine TranslationNMTSemantic Similarity+2

Stepwise Credit Assignment for GRPO on Flow-Matching Models

2026-03-30 · Yash Savani, Branislav Kveton, Yuchen Liu, Yilin Wang 외 arxiv

Flow-GRPO successfully applies reinforcement learning to flow models, but uses uniform credit assignment across all steps. This ignores the temporal structure of diffusion generation: early steps determine composition an…

Reinforcement Learning

MACCA: Offline Multi-agent Reinforcement Learning with Causal Credit Assignment

2023-12-06 · Ziyan Wang, Yali Du, Yudi Zhang, Meng Fang 외

Offline Multi-agent Reinforcement Learning (MARL) is valuable in scenarios where online interaction is impractical or risky. While independent learning in MARL offers flexibility and scalability, accurately assigning cre…

Multi-agent Reinforcement Learningreinforcement-learningReinforcement Learning