対戦型ゲームにおける戦略多様性についてのStGA法を用いた自動分析手法の提案とその評価 Proposal and Evaluation of an Automatic Analysis Method Using StGA for Strategic Diversity in Action Games

この論文にアクセスする

この論文をさがす

著者

    • 山本 界人 Kaito Yamamoto
    • 立命館大学情報理工学部|現在,立命館大学大学院情報理工学研究科 College of Information Science & Engineering, Ritsumeikan University | Presently with Graduate School of Information Science & Engineering, Ritsumeikan University
    • 水野 竣介 Shunsuke Mizuno
    • 立命館大学情報理工学部|現在,株式会社ディンプス College of Information Science & Engineering, Ritsumeikan University | Presently with Dimps Corporation

抄録

本稿では,対戦型ゲームにおける戦略多様性の観点からゲームバランスの分析を自動的に行う手法を提案し,その有用性を評価した.ゲーム開発におけるゲームバランスの分析,調整は面白いゲームを作成するためには不可欠な要素である.一方で,その分析,調整のプロセスには時間的コストがかかる.開発期間に限りがあるゲーム開発の現場では,十分にゲームバランスの分析,調整を行うことができない場合も少なくない.このため,ゲームバランスを自動的に分析する手法が必要とされている.既存のゲームバランスの自動分析手法は,1つの状況を分析するために大きな計算時間を必要とする,もしくは事前に人間がゲームタイトルに依存する専門的な知識を必要とするものだった.このため,人間が知識を獲得していない,かつ多くの状況が存在するゲームへと適用する場合には大きな計算時間を必要とした.そこで,本稿では,Stochastic Genetic Algorithm(StGA)を用いて,専門的な知識なしに多くの状況を持つゲームを分析する手法を提案する.国際AI大会のプラットフォームとして利用されているFightingICEを対象にした実験から,本手法の有用性を確かめた.In this paper, we propose and evaluate an automatic analysis method for game balance from the perspective of strategic diversity in action games. Analysis and adjustment of game balance in game development are essential elements in order to create an interesting game. However, time cost is required for such analysis and adjustment. In typical game development, often with limited development time, it is not possible to adjust and analyze game balance completely. Therefore, an automatic method for analyzing the game balance is required. So far, existing automatic analysis methods need significant computational time to analyze a situation, or require specialized knowledge obtained by humans based on the game title in advance. Therefore, a large amount of computational time is required when applying those systems to games with many situations, of which knowledge is not yet acquired by humans. In this paper, we propose a method that uses Stochastic Genetic Algorithm (StGA) for analyzing that sort of games. From experiments using FightingICE, which is used as the platform for an international AI tournament, the effectiveness of our method is confirmed.

In this paper, we propose and evaluate an automatic analysis method for game balance from the perspective of strategic diversity in action games. Analysis and adjustment of game balance in game development are essential elements in order to create an interesting game. However, time cost is required for such analysis and adjustment. In typical game development, often with limited development time, it is not possible to adjust and analyze game balance completely. Therefore, an automatic method for analyzing the game balance is required. So far, existing automatic analysis methods need significant computational time to analyze a situation, or require specialized knowledge obtained by humans based on the game title in advance. Therefore, a large amount of computational time is required when applying those systems to games with many situations, of which knowledge is not yet acquired by humans. In this paper, we propose a method that uses Stochastic Genetic Algorithm (StGA) for analyzing that sort of games. From experiments using FightingICE, which is used as the platform for an international AI tournament, the effectiveness of our method is confirmed.

収録刊行物

  • 情報処理学会論文誌

    情報処理学会論文誌 55(11), 2328-2335, 2014-11-15

    一般社団法人情報処理学会

各種コード

  • NII論文ID(NAID)
    110009843037
  • NII書誌ID(NCID)
    AN00116647
  • 本文言語コード
    JPN
  • 資料種別
    Journal Article
  • ISSN
    1882-7764
  • データ提供元
    NII-ELS  IPSJ 
ページトップへ