Krzysztof Krawiec


Home

Research:

edit SideBar

Recent developments cast doubts on the effectiveness of coevolutionary learning in interactive domains. A simple evolution with fitness evaluation based on games with random strategies has been found to generalize better than competitive coevolution. In an attempt to investigate this phenomenon, we analyze the utility of random opponents for one and two-population competitive coevolution applied to learning strategies for the game of Othello. We show that if coevolution uses two-population setup and engages also random opponents, it is capable of producing equally good strategies as evolution with random sampling for the expected utility performance measure. To investigate the differences between analyzed methods, we introduce performance profile, a tool that measures the player's performance against opponents of various strength. The profiles reveal that evolution with random sampling produces players coping well with mediocre opponents, but playing relatively poorly against stronger ones. This finding explains why in the round-robin tournament, evolution with random sampling is one of the worst methods from all those considered in this study.

@INPROCEEDINGS { Jaskowski:2013:ICR,
    AUTHOR = { Jaśkowski, Wojciech and Liskowski, Paweł and Szubert, Marcin and Krawiec, Krzysztof },
    TITLE = { Improving coevolution by random sampling },
    BOOKTITLE = { Proceeding of the fifteenth annual conference on Genetic and evolutionary computation conference },
    SERIES = { GECCO '13 },
    YEAR = { 2013 },
    ISBN = { 978-1-4503-1963-8 },
    LOCATION = { Amsterdam, The Netherlands },
    PAGES = { 1141--1148 },
    NUMPAGES = { 8 },
    URL = { http://doi.acm.org/10.1145/2463372.2463512 },
    DOI = { 10.1145/2463372.2463512 },
    ACMID = { 2463512 },
    PUBLISHER = { ACM },
    ADDRESS = { New York, NY, USA },
    KEYWORDS = { competitive coevolution, maximization of expected utility, othello, performance profile, solution concepts, strategy learning },
    ABSTRACT = { Recent developments cast doubts on the effectiveness of coevolutionary learning in interactive domains. A simple evolution with fitness evaluation based on games with random strategies has been found to generalize better than competitive coevolution. In an attempt to investigate this phenomenon, we analyze the utility of random opponents for one and two-population competitive coevolution applied to learning strategies for the game of Othello. We show that if coevolution uses two-population setup and engages also random opponents, it is capable of producing equally good strategies as evolution with random sampling for the expected utility performance measure. To investigate the differences between analyzed methods, we introduce performance profile, a tool that measures the player's performance against opponents of various strength. The profiles reveal that evolution with random sampling produces players coping well with mediocre opponents, but playing relatively poorly against stronger ones. This finding explains why in the round-robin tournament, evolution with random sampling is one of the worst methods from all those considered in this study. },
}


Powered by PmWiki