Combinatorial Optimization Method Based on Hierarchical Structure in Solution Space Utilizing Stochastic Neighborhood Selection

カテゴリ: 部門大会

論文No: SS2-3

グループ名: 【C】2023年電気学会電子・情報・システム部門大会

発行日: 2023/08/23

タイトル(英語): Combinatorial Optimization Method Based on Hierarchical Structure in Solution Space Utilizing Stochastic Neighborhood Selection

著者名: 仲田 圭吾（東京都立大学）,関井 大輔（沖電気工業）,田村 健一（東京都立大学）,安田 恵一郎（東京都立大学）

著者名(英語): Keigo Nakada (Tokyo Metropolitan University),Daisuke Sekii (Oki Electric Industry),Kenichi Tamura (Tokyo Metropolitan University),Keiichiro Yasuda (Tokyo Metropolitan University)

キーワード: 組合せ最適化|メタヒューリスティクス|多点探索|局所探索法|確率論的過程|Combinational Optimization|Metaheuristics|Multi-point Search|Local Search|Stochastic Process

要約(日本語): As systems become larger and more complex, real-world problems require quasi-optimal solutions be obtained in practical time. Meta-heuristics have attention as a framework for methods that seek quasi-optimal solutions. We focus on the degeneracy in the combinatorial optimization method with search strategy based on hierarchical interpretation of solution space, which has high performance compared to existing basic methods. The degeneracy is an issue in which search points follow the same path. It wastes computational resources and weakens the interaction between search points. A new stochastic process is introduced to improve search performance by dealing with the degeneracy. The occurrence of the degeneracy and search performance are compared and verified with the original method.