作成者 |
|
|
|
本文言語 |
|
出版者 |
|
|
発行日 |
|
収録物名 |
|
開始ページ |
|
終了ページ |
|
出版タイプ |
|
アクセス権 |
|
関連DOI |
|
|
関連URI |
|
|
関連情報 |
|
|
概要 |
We make a Star Trek game player model with binary and fuzzy rules, parameterize the rules, and tune them using differential evolution (DE) and genetic algorithms (GA). First, we built a human player m...odel player using binary and fuzzy rules that can fight with Star Trek game in a computer. Second, we parameterize the rules and apply DE and GA to them. This task was so difficult that DE and GA could not find optimum parameters that make the player model winner. To solve this situation, we apply new idea that changes the search environment gradually during the evolution process to find better solutions. We start from the simplest conditions of Star Trek game and gradually make them difficult according to the number of winnings of the human player model. This co-evolution-like strategy could find better rules parameters and the player model became stronger, while just applying DE and GA did not converge to its solution.続きを見る
|