Constructing Search Algorithms Taking Account of Global-Multimodality and Many-Objectiveness

2016 Fiscal Year Final Research Report

• Project/Area Number: 26330272
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Soft computing
• Research Institution: Tokyo Institute of Technology
• Principal Investigator: Ono Isao 東京工業大学, 情報理工学院, 准教授 (00304551)
• Project Period (FY): 2014-04-01 – 2017-03-31
• Keywords: 進化計算 / 大域的最適化 / 多目的最適化 / 大域的多峰性 / 連続関数最適化 / 自然進化戦略 / 有望領域囲い込み法 / AWA

Outline of Final Research Achievements

In this study, we proposed new search methods for efficiently finding good approximate solutions in globally-multimodal optimization problems and new ones for efficiently searching for good approximate solution sets in many-objective optimization problems. From the viewpoint of searching for good approximate solutions in the globally-multimodal search space, we proposed new methods to discover big-valleys with the optimal solutions and to search for the best solution in each big-valley. We showed that the proposed methods showed better performance than conventional ones on benchmark problems and real-world ones. From the viewpoint of searching for good approximate solution sets in many-objective optimization problems, we proposed new multi-start search methods based on scalarization that are excellent in terms of the coverage. We demonstrated that the proposed methods outperformed conventional ones on benchmark problems.

Free Research Field

進化計算