Self-generation of Optimization criteria on Evolutionary Computation for Computationally-expensive optimization problems
| Project/Area Number |
18K18123
|
|---|
| Research Category |
Grant-in-Aid for Early-Career Scientists
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Basic Section 61040:Soft computing-related
|
|---|
| Research Institution | Yokohama National University |
|---|
Principal Investigator |
Nakata Masaya 横浜国立大学, 大学院工学研究院, 准教授 (00781072)
|
|---|
| Project Period (FY) |
2018-04-01 – 2020-03-31
|
| Project Status |
Completed (Fiscal Year 2019)
|
| Budget Amount *help |
¥4,160,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥960,000)
Fiscal Year 2019: ¥1,820,000 (Direct Cost: ¥1,400,000、Indirect Cost: ¥420,000)
Fiscal Year 2018: ¥2,340,000 (Direct Cost: ¥1,800,000、Indirect Cost: ¥540,000)
|
|---|
| Keywords | 進化計算 / メタヒューリスティックス / 機械学習 / 進化的機械学習 / 最適化 |
|---|
| Outline of Final Research Achievements |
This project proposed an evolutionary optimization technique that identifies and then uses useful optimization criteria based on solution structure and solution variable dependency. Generated optimization criteria require no additional evaluations of evaluations, and thus the proposed method can be suitable for computationally expensive optimization problems. While the proposed method was initially designed for a single-objective optimization problem, beyond this goal, this project extended our methodology to large-scale optimization problems and multi-objective optimization problems. Accordingly, this project provides the following contributions. Firstly, optimizing based on an extracted solution structure of good solutions can enhance an optimization performance especially in single-objective problems including large-scape optimization problems. Secondly, an SVM-based optimization criterion which predicts good solutions is suitable for multi-objective optimization problems.
|
| Academic Significance and Societal Importance of the Research Achievements |
工学設計における最適化問題の多くは、１つの解評価に数時間から数日必要となる高計算コストな最適化問題に属する場合が多い。この場合、可能な限り少ない解評価で良好な解の導出を求めることが重要となる。本研究は、専門的な知識がなくとも利用できる使い勝手が良い進化的最適化法に、解評価を必要とせずに最適化を促進する方法論とその実装方法明らかにすることで、高計算コストな問題に特化した効率の良い手法を構築した点に意義がある。加えて、この手法は、実最適化問題で頻出する大規模問題、多目的最適化問題にも対応できることを示した。
|
Report
(3 results)
Research Products
(10 results)
