文献类型：期刊文献

英文题名：Improving value function decomposition in cooperative multi-agent reinforcement learning

作者：Qin, Hao[1];Hong, Chen[2,3]

第一作者：Qin, Hao

通讯作者：Hong, C[1]

机构：[1]Beijing Union Univ, Coll Smart City, Beijing 100101, Peoples R China;[2]Beijing Union Univ, Multiagent Syst Res Ctr, Beijing 100101, Peoples R China;[3]Beijing Union Univ, Coll Robot, Beijing 100101, Peoples R China

第一机构：北京联合大学

通讯机构：[1]corresponding author), Beijing Union Univ, Multiagent Syst Res Ctr, Beijing 100101, Peoples R China.|[11417]北京联合大学;

年份：2026

卷号：703

外文期刊名：NEUROCOMPUTING

收录：;EI(收录号：20263221275342);Scopus(收录号：2-s2.0-105046863996);WOS:【SCI-EXPANDED(收录号:WOS:001847449600001)】；

基金：This paper is supported by the National Science Foundation for Young Scientists of China (Grant No. 52502397) .

语种：英文

外文关键词：Cooperative multi-agent reinforcement; learning; Value decomposition; Dynamic coordination graph; Graph attention networks; Residual utility enhancement

摘要：Value decomposition methods have achieved strong performance in cooperative multi-agent reinforcement learning (MARL) under the centralized training with decentralized execution paradigm. However, effectively incorporating explicit, sparse, and time-varying coordination structures into monotonic value decomposition remains challenging. Although recent graph-based MARL methods have explored dynamic and group-aware in teraction modeling, it is still non-trivial to construct lightweight and interpretable coordination graphs from local observations while preserving stable utility learning and decentralized execution. To address this issue, we propose the Dynamic Graph Attention Mixing Network (DGAT-MIX), a value decomposition framework for ex plicit dynamic coordination modeling. DGAT-MIX constructs an observation-driven coordination graph from local visibility and distance cues and performs masked multi-head graph attention over the resulting graph. The graph-enhanced representations are then injected into individual utilities through a residual coordination signal before monotonic mixing. This design introduces a sparse and interpretable relational inductive bias while retaining the original utility pathway for stable optimization. Experiments on MPE and SMACv2, together with additional anal yses on selected SMAC scenarios, show that DGAT-MIX achieves competitive or superior performance compared with representative baselines, especially in heterogeneous, asymmetric, and dynamically changing cooperative tasks.