Formation of association structures based on reciprocity and their performance in allocation problems

Yuki Miyashita, Masashi Hayano, Toshiharu Sugawara

    研究成果: Conference contribution

    2 引用 (Scopus)

    抄録

    We describe the reciprocal agents that build virtual associations in accordance with past cooperative work in a bottom-up manner and that allocate tasks or resources preferentially to agents in the same associations in busy large-scale distributed environments. Models of multi-agent systems (MAS) are often used to express tasks that are done by teams of cooperative agents, so how each subtask is allocated to appropriate agents is a central issue. Particularly in busy environments where multiple tasks are requested simultaneously and continuously, simple allocation methods in self-interested agents result in conflicts, meaning that these methods attempt to allocate multiple tasks to one or a few capable agents. Thus, the system’s performance degrades. To avoid such conflicts, we introduce reciprocal agents that cooperate with specific agents that have excellent mutual experience of cooperation. They then autonomously build associations in which they try to form teams for new incoming tasks. We introduce the N-agent team formation (TF) game, an abstract expression of allocating problems in MAS by eliminating unnecessary and complicated task and agent specifications, thereby identifying the fundamental mechanism to facilitate and maintain associations. We experimentally show that reciprocal agents can considerably improve performance by reducing the number of conflicts in N-agent TF games with different N values by establishing association structures. We also investigate how learning parameters to decide reciprocity affect association structures and which structure can achieve efficient allocations.

    元の言語English
    ホスト出版物のタイトルCoordination, Organizations, Institutions, and Normes in Agent Systems XI - COIN 2015 International Workshops COIN@AAMAS, Revised Selected Papers
    出版者Springer Verlag
    ページ262-281
    ページ数20
    9628
    ISBN(印刷物)9783319426907
    DOI
    出版物ステータスPublished - 2016
    イベントInternational Conference on Coordination, Organisations, Institutions and Norms in Agent Systems, 2015 - Istanbul, Turkey
    継続期間: 2015 5 42015 5 4

    出版物シリーズ

    名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    9628
    ISSN(印刷物)03029743
    ISSN(電子版)16113349

    Other

    OtherInternational Conference on Coordination, Organisations, Institutions and Norms in Agent Systems, 2015
    Turkey
    Istanbul
    期間15/5/415/5/4

    Fingerprint

    Reciprocity
    Multi agent systems
    Multi-agent Systems
    Game
    Parameter Learning
    Distributed Environment
    Bottom-up
    System Performance
    Specifications
    Express
    Specification

    ASJC Scopus subject areas

    • Computer Science(all)
    • Theoretical Computer Science

    これを引用

    Miyashita, Y., Hayano, M., & Sugawara, T. (2016). Formation of association structures based on reciprocity and their performance in allocation problems. : Coordination, Organizations, Institutions, and Normes in Agent Systems XI - COIN 2015 International Workshops COIN@AAMAS, Revised Selected Papers (巻 9628, pp. 262-281). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); 巻数 9628). Springer Verlag. https://doi.org/10.1007/978-3-319-42691-4_15

    Formation of association structures based on reciprocity and their performance in allocation problems. / Miyashita, Yuki; Hayano, Masashi; Sugawara, Toshiharu.

    Coordination, Organizations, Institutions, and Normes in Agent Systems XI - COIN 2015 International Workshops COIN@AAMAS, Revised Selected Papers. 巻 9628 Springer Verlag, 2016. p. 262-281 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); 巻 9628).

    研究成果: Conference contribution

    Miyashita, Y, Hayano, M & Sugawara, T 2016, Formation of association structures based on reciprocity and their performance in allocation problems. : Coordination, Organizations, Institutions, and Normes in Agent Systems XI - COIN 2015 International Workshops COIN@AAMAS, Revised Selected Papers. 巻. 9628, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 巻. 9628, Springer Verlag, pp. 262-281, International Conference on Coordination, Organisations, Institutions and Norms in Agent Systems, 2015, Istanbul, Turkey, 15/5/4. https://doi.org/10.1007/978-3-319-42691-4_15
    Miyashita Y, Hayano M, Sugawara T. Formation of association structures based on reciprocity and their performance in allocation problems. : Coordination, Organizations, Institutions, and Normes in Agent Systems XI - COIN 2015 International Workshops COIN@AAMAS, Revised Selected Papers. 巻 9628. Springer Verlag. 2016. p. 262-281. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-42691-4_15
    Miyashita, Yuki ; Hayano, Masashi ; Sugawara, Toshiharu. / Formation of association structures based on reciprocity and their performance in allocation problems. Coordination, Organizations, Institutions, and Normes in Agent Systems XI - COIN 2015 International Workshops COIN@AAMAS, Revised Selected Papers. 巻 9628 Springer Verlag, 2016. pp. 262-281 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
    @inproceedings{5d7b921efe464c6881ea5ba5bc0cc201,
    title = "Formation of association structures based on reciprocity and their performance in allocation problems",
    abstract = "We describe the reciprocal agents that build virtual associations in accordance with past cooperative work in a bottom-up manner and that allocate tasks or resources preferentially to agents in the same associations in busy large-scale distributed environments. Models of multi-agent systems (MAS) are often used to express tasks that are done by teams of cooperative agents, so how each subtask is allocated to appropriate agents is a central issue. Particularly in busy environments where multiple tasks are requested simultaneously and continuously, simple allocation methods in self-interested agents result in conflicts, meaning that these methods attempt to allocate multiple tasks to one or a few capable agents. Thus, the system’s performance degrades. To avoid such conflicts, we introduce reciprocal agents that cooperate with specific agents that have excellent mutual experience of cooperation. They then autonomously build associations in which they try to form teams for new incoming tasks. We introduce the N-agent team formation (TF) game, an abstract expression of allocating problems in MAS by eliminating unnecessary and complicated task and agent specifications, thereby identifying the fundamental mechanism to facilitate and maintain associations. We experimentally show that reciprocal agents can considerably improve performance by reducing the number of conflicts in N-agent TF games with different N values by establishing association structures. We also investigate how learning parameters to decide reciprocity affect association structures and which structure can achieve efficient allocations.",
    author = "Yuki Miyashita and Masashi Hayano and Toshiharu Sugawara",
    year = "2016",
    doi = "10.1007/978-3-319-42691-4_15",
    language = "English",
    isbn = "9783319426907",
    volume = "9628",
    series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
    publisher = "Springer Verlag",
    pages = "262--281",
    booktitle = "Coordination, Organizations, Institutions, and Normes in Agent Systems XI - COIN 2015 International Workshops COIN@AAMAS, Revised Selected Papers",
    address = "Germany",

    }

    TY - GEN

    T1 - Formation of association structures based on reciprocity and their performance in allocation problems

    AU - Miyashita, Yuki

    AU - Hayano, Masashi

    AU - Sugawara, Toshiharu

    PY - 2016

    Y1 - 2016

    N2 - We describe the reciprocal agents that build virtual associations in accordance with past cooperative work in a bottom-up manner and that allocate tasks or resources preferentially to agents in the same associations in busy large-scale distributed environments. Models of multi-agent systems (MAS) are often used to express tasks that are done by teams of cooperative agents, so how each subtask is allocated to appropriate agents is a central issue. Particularly in busy environments where multiple tasks are requested simultaneously and continuously, simple allocation methods in self-interested agents result in conflicts, meaning that these methods attempt to allocate multiple tasks to one or a few capable agents. Thus, the system’s performance degrades. To avoid such conflicts, we introduce reciprocal agents that cooperate with specific agents that have excellent mutual experience of cooperation. They then autonomously build associations in which they try to form teams for new incoming tasks. We introduce the N-agent team formation (TF) game, an abstract expression of allocating problems in MAS by eliminating unnecessary and complicated task and agent specifications, thereby identifying the fundamental mechanism to facilitate and maintain associations. We experimentally show that reciprocal agents can considerably improve performance by reducing the number of conflicts in N-agent TF games with different N values by establishing association structures. We also investigate how learning parameters to decide reciprocity affect association structures and which structure can achieve efficient allocations.

    AB - We describe the reciprocal agents that build virtual associations in accordance with past cooperative work in a bottom-up manner and that allocate tasks or resources preferentially to agents in the same associations in busy large-scale distributed environments. Models of multi-agent systems (MAS) are often used to express tasks that are done by teams of cooperative agents, so how each subtask is allocated to appropriate agents is a central issue. Particularly in busy environments where multiple tasks are requested simultaneously and continuously, simple allocation methods in self-interested agents result in conflicts, meaning that these methods attempt to allocate multiple tasks to one or a few capable agents. Thus, the system’s performance degrades. To avoid such conflicts, we introduce reciprocal agents that cooperate with specific agents that have excellent mutual experience of cooperation. They then autonomously build associations in which they try to form teams for new incoming tasks. We introduce the N-agent team formation (TF) game, an abstract expression of allocating problems in MAS by eliminating unnecessary and complicated task and agent specifications, thereby identifying the fundamental mechanism to facilitate and maintain associations. We experimentally show that reciprocal agents can considerably improve performance by reducing the number of conflicts in N-agent TF games with different N values by establishing association structures. We also investigate how learning parameters to decide reciprocity affect association structures and which structure can achieve efficient allocations.

    UR - http://www.scopus.com/inward/record.url?scp=84978811042&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84978811042&partnerID=8YFLogxK

    U2 - 10.1007/978-3-319-42691-4_15

    DO - 10.1007/978-3-319-42691-4_15

    M3 - Conference contribution

    AN - SCOPUS:84978811042

    SN - 9783319426907

    VL - 9628

    T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

    SP - 262

    EP - 281

    BT - Coordination, Organizations, Institutions, and Normes in Agent Systems XI - COIN 2015 International Workshops COIN@AAMAS, Revised Selected Papers

    PB - Springer Verlag

    ER -