Interpretability for Conditional Coordinated Behavior in Multi-Agent Reinforcement Learning

被引:0
|
作者
Motokawa, Yoshinari [1 ]
Sugawara, Toshiharu [1 ]
机构
[1] Waseda Univ, Dept Comp Sci, Tokyo, Japan
关键词
Multi-agent deep reinforcement learning; XRL; Distributed system; Attentional mechanism; Coordination; Cooperation;
D O I
10.1109/IJCNN54540.2023.10191825
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We propose a model-free reinforcement learning architecture, called distributed attentional actor architecture after conditional attention (DA6-X), to provide better interpretability of conditional coordinated behaviors. The underlying principle involves reusing the saliency vector, which represents the conditional states of the environment, such as the global position of agents. Hence, agents with DA6-X flexibility built into their policy exhibit superior performance by considering the additional information in the conditional states during the decision-making process. The effectiveness of the proposed method was experimentally evaluated by comparing it with conventional methods in an objects collection game. By visualizing the attention weights from DA6-X, we confirmed that agents successfully learn situation-dependent coordinated behaviors by correctly identifying various conditional states, leading to improved interpretability of agents along with superior performance.
引用
收藏
页数:8
相关论文
共 50 条
  • [41] Learning structured communication for multi-agent reinforcement learning
    Sheng, Junjie
    Wang, Xiangfeng
    Jin, Bo
    Yan, Junchi
    Li, Wenhao
    Chang, Tsung-Hui
    Wang, Jun
    Zha, Hongyuan
    [J]. AUTONOMOUS AGENTS AND MULTI-AGENT SYSTEMS, 2022, 36 (02)
  • [42] Learning structured communication for multi-agent reinforcement learning
    Junjie Sheng
    Xiangfeng Wang
    Bo Jin
    Junchi Yan
    Wenhao Li
    Tsung-Hui Chang
    Jun Wang
    Hongyuan Zha
    [J]. Autonomous Agents and Multi-Agent Systems, 2022, 36
  • [43] Generalized learning automata for multi-agent reinforcement learning
    De Hauwere, Yann-Michael
    Vrancx, Peter
    Nowe, Ann
    [J]. AI COMMUNICATIONS, 2010, 23 (04) : 311 - 324
  • [44] Multi-agent deep reinforcement learning: a survey
    Sven Gronauer
    Klaus Diepold
    [J]. Artificial Intelligence Review, 2022, 55 : 895 - 943
  • [45] Reinforcement learning based on multi-agent in RoboCup
    Zhang, W
    Li, JG
    Ruan, XG
    [J]. ADVANCES IN INTELLIGENT COMPUTING, PT 1, PROCEEDINGS, 2005, 3644 : 967 - 975
  • [46] Multi-Agent Adversarial Inverse Reinforcement Learning
    Yu, Lantao
    Song, Jiaming
    Ermon, Stefano
    [J]. INTERNATIONAL CONFERENCE ON MACHINE LEARNING, VOL 97, 2019, 97
  • [47] Parallel and distributed multi-agent reinforcement learning
    Kaya, M
    Arslan, A
    [J]. PROCEEDINGS OF THE EIGHTH INTERNATIONAL CONFERENCE ON PARALLEL AND DISTRIBUTED SYSTEMS, 2001, : 437 - 441
  • [48] Coding for Distributed Multi-Agent Reinforcement Learning
    Wang, Baoqian
    Xie, Junfei
    Atanasov, Nikolay
    [J]. 2021 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION (ICRA 2021), 2021, : 10625 - 10631
  • [49] Multi-agent reinforcement learning for character control
    Li, Cheng
    Fussell, Levi
    Komura, Taku
    [J]. VISUAL COMPUTER, 2021, 37 (12): : 3115 - 3123
  • [50] Multi-agent Reinforcement Learning for Service Composition
    Lei, Yu
    Yu, Philip S.
    [J]. PROCEEDINGS 2016 IEEE INTERNATIONAL CONFERENCE ON SERVICES COMPUTING (SCC 2016), 2016, : 790 - 793