Modifying MCTS for human-like general video game playing

Ahmed Khalifa, Aaron Isaksen, Julian Togelius, Andrew Nealen

    Research output: Contribution to journalArticle

    Abstract

    We address the problem of making general video game playing agents play in a human-like manner. To this end, we introduce several modifications of the UCT formula used in Monte Carlo Tree Search that biases action selection towards repeating the current action, making pauses, and limiting rapid switching between actions. Playtraces of human players are used to model their propensity for repeated actions; this model is used for biasing the UCT formula. Experiments show that our modified MCTS agent, called BoT, plays quantitatively similar to human players as measured by the distribution of repeated actions. A survey of human observers reveals that the agent exhibits human-like playing style in some games but not others.

    Original languageEnglish (US)
    Pages (from-to)2514-2520
    Number of pages7
    JournalIJCAI International Joint Conference on Artificial Intelligence
    Volume2016-January
    StatePublished - 2016

    Fingerprint

    Experiments

    ASJC Scopus subject areas

    • Artificial Intelligence

    Cite this

    Khalifa, A., Isaksen, A., Togelius, J., & Nealen, A. (2016). Modifying MCTS for human-like general video game playing. IJCAI International Joint Conference on Artificial Intelligence, 2016-January, 2514-2520.

    Modifying MCTS for human-like general video game playing. / Khalifa, Ahmed; Isaksen, Aaron; Togelius, Julian; Nealen, Andrew.

    In: IJCAI International Joint Conference on Artificial Intelligence, Vol. 2016-January, 2016, p. 2514-2520.

    Research output: Contribution to journalArticle

    Khalifa, A, Isaksen, A, Togelius, J & Nealen, A 2016, 'Modifying MCTS for human-like general video game playing', IJCAI International Joint Conference on Artificial Intelligence, vol. 2016-January, pp. 2514-2520.
    Khalifa, Ahmed ; Isaksen, Aaron ; Togelius, Julian ; Nealen, Andrew. / Modifying MCTS for human-like general video game playing. In: IJCAI International Joint Conference on Artificial Intelligence. 2016 ; Vol. 2016-January. pp. 2514-2520.
    @article{dbf1798caaa249e48e734db724154991,
    title = "Modifying MCTS for human-like general video game playing",
    abstract = "We address the problem of making general video game playing agents play in a human-like manner. To this end, we introduce several modifications of the UCT formula used in Monte Carlo Tree Search that biases action selection towards repeating the current action, making pauses, and limiting rapid switching between actions. Playtraces of human players are used to model their propensity for repeated actions; this model is used for biasing the UCT formula. Experiments show that our modified MCTS agent, called BoT, plays quantitatively similar to human players as measured by the distribution of repeated actions. A survey of human observers reveals that the agent exhibits human-like playing style in some games but not others.",
    author = "Ahmed Khalifa and Aaron Isaksen and Julian Togelius and Andrew Nealen",
    year = "2016",
    language = "English (US)",
    volume = "2016-January",
    pages = "2514--2520",
    journal = "IJCAI International Joint Conference on Artificial Intelligence",
    issn = "1045-0823",

    }

    TY - JOUR

    T1 - Modifying MCTS for human-like general video game playing

    AU - Khalifa, Ahmed

    AU - Isaksen, Aaron

    AU - Togelius, Julian

    AU - Nealen, Andrew

    PY - 2016

    Y1 - 2016

    N2 - We address the problem of making general video game playing agents play in a human-like manner. To this end, we introduce several modifications of the UCT formula used in Monte Carlo Tree Search that biases action selection towards repeating the current action, making pauses, and limiting rapid switching between actions. Playtraces of human players are used to model their propensity for repeated actions; this model is used for biasing the UCT formula. Experiments show that our modified MCTS agent, called BoT, plays quantitatively similar to human players as measured by the distribution of repeated actions. A survey of human observers reveals that the agent exhibits human-like playing style in some games but not others.

    AB - We address the problem of making general video game playing agents play in a human-like manner. To this end, we introduce several modifications of the UCT formula used in Monte Carlo Tree Search that biases action selection towards repeating the current action, making pauses, and limiting rapid switching between actions. Playtraces of human players are used to model their propensity for repeated actions; this model is used for biasing the UCT formula. Experiments show that our modified MCTS agent, called BoT, plays quantitatively similar to human players as measured by the distribution of repeated actions. A survey of human observers reveals that the agent exhibits human-like playing style in some games but not others.

    UR - http://www.scopus.com/inward/record.url?scp=85006153505&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=85006153505&partnerID=8YFLogxK

    M3 - Article

    AN - SCOPUS:85006153505

    VL - 2016-January

    SP - 2514

    EP - 2520

    JO - IJCAI International Joint Conference on Artificial Intelligence

    JF - IJCAI International Joint Conference on Artificial Intelligence

    SN - 1045-0823

    ER -