| 講演名 | 2020-05-15 Spy Algorithm for Solving Continuous Optimization Problem Dhidhi Pambudi(山口大), Masaki Kawamura(山口大), |
|---|---|
| PDFダウンロードページ |  PDFダウンロードページへ |
| 抄録(和) | We proposed the Spy algorithm for solving continuous optimization problems on multimodal functions. The algorithm is one of population-based metaheuristics and is inspired by spy-ring (a cooperated group of spies). The multimodal function may have many local optimums and some of them can be close to the global optimum. Some optimization methods can be trapped in these local optimum and fail to reach the global optimum. Moreover, multimodal function can have many flat plateaus and valleys that make it much harder to solve. The Spy algorithm accommodates exploitation and exploration search in a simple concept and simple operation that make it easy for tuning to get a good result. In order to check the performance of the proposed method, we applied it to some benchmarking tests, which are Ackley, Schwefel, Rastrigin, Dixon-Price, and Michalewicz function on dimension size 3, 5, 10, 15, and 20. We compare the Spy algorithm against Cuckoo Search with and without Levy flight and FFA. Based on the result, the average of error on Ackley, Dixon-Price, and Michalewicz function regardless of dimension size for Spy Algorithm, CS non-Levy, CS Levy, and FFA are 0.48, 18.12, 24.79, and 67921.71. All algorithm was weak on Rastrigin function and very weak on Schwefel function, but the Spy Algorithm still overcame all other compared algorithm. The average computation time regardless of the test functions on all dimension size for Spy Algorithm, CS non-Levy, CS Levy, and FFA are 0.46s, 0.23s, 1.02s, and 5.09s. The Spy algorithm was fast enough and had the best accuracy. |
| 抄録(英) | We proposed the Spy algorithm for solving continuous optimization problems on multimodal functions. The algorithm is one of population-based metaheuristics and is inspired by spy-ring (a cooperated group of spies). The multimodal function may have many local optimums and some of them can be close to the global optimum. Some optimization methods can be trapped in these local optimum and fail to reach the global optimum. Moreover, multimodal function can have many flat plateaus and valleys that make it much harder to solve. The Spy algorithm accommodates exploitation and exploration search in a simple concept and simple operation that make it easy for tuning to get a good result. In order to check the performance of the proposed method, we applied it to some benchmarking tests, which are Ackley, Schwefel, Rastrigin, Dixon-Price, and Michalewicz function on dimension size 3, 5, 10, 15, and 20. We compare the Spy algorithm against Cuckoo Search with and without Levy flight and FFA. Based on the result, the average of error on Ackley, Dixon-Price, and Michalewicz function regardless of dimension size for Spy Algorithm, CS non-Levy, CS Levy, and FFA are 0.48, 18.12, 24.79, and 67921.71. All algorithm was weak on Rastrigin function and very weak on Schwefel function, but the Spy Algorithm still overcame all other compared algorithm. The average computation time regardless of the test functions on all dimension size for Spy Algorithm, CS non-Levy, CS Levy, and FFA are 0.46s, 0.23s, 1.02s, and 5.09s. The Spy algorithm was fast enough and had the best accuracy. |
| キーワード(和) | optimization / continuous / metaheuristic |
| キーワード(英) | optimization / continuous / metaheuristic |
| 資料番号 | NLP2020-5 |
| 発行日 | 2020-05-08 (NLP) |
| 研究会情報 | |
| 研究会 | NLP |
|---|---|
| 開催期間 | 2020/5/15(から1日開催) |
| 開催地（和） | 山口大学（常盤キャンパス） |
| 開催地（英） | Yamaguchi University (Tokiwa campus) |
| テーマ（和） | 一般 |
| テーマ（英） | |
| 委員長氏名（和） | 黒川 弘章(東京工科大) |
| 委員長氏名（英） | Hiroaki Kurokawa(Tokyo Univ. of Tech.) |
| 副委員長氏名（和） | 夏目 季代久(九工大) |
| 副委員長氏名（英） | Kiyohisa Natsume(Kyushu Inst. of Tech.) |
| 幹事氏名（和） | 木村 貴幸(日本工大) / 立野 勝巳(九工大) |
| 幹事氏名（英） | Takayuki Kimura(Nippon Inst. of Tech.) / Katsumi Tateno(Kyushu Inst. of Tech.) |
| 幹事補佐氏名（和） | 島田 裕(埼玉大) / 佐村 俊和(山口大) |
| 幹事補佐氏名（英） | Yutaka Shimada(Saitama Univ.) / Toshikaza Samura(Yamaguchi Univ.) |
| 講演論文情報詳細 | |
| 申込み研究会 | Technical Committee on Nonlinear Problems |
|---|---|
| 本文の言語 | ENG |
| タイトル（和） | |
| サブタイトル（和） | |
| タイトル（英） | Spy Algorithm for Solving Continuous Optimization Problem |
| サブタイトル（和） | |
| キーワード(1)（和/英） | optimization / optimization |
| キーワード(2)（和/英） | continuous / continuous |
| キーワード(3)（和/英） | metaheuristic / metaheuristic |
| 第 1 著者 氏名（和/英） | Dhidhi Pambudi / Dhidhi Pambudi |
| 第 1 著者 所属（和/英） | Yamaguchi University(略称：山口大) Yamaguchi University(略称：Yamaguchi Univ.) |
| 第 2 著者 氏名（和/英） | Masaki Kawamura / Masaki Kawamura |
| 第 2 著者 所属（和/英） | Yamaguchi University(略称：山口大) Yamaguchi University(略称：Yamaguchi Univ.) |
| 発表年月日 | 2020-05-15 |
| 資料番号 | NLP2020-5 |
| 巻番号（vol） | vol.120 |
| 号番号（no） | NLP-26 |
| ページ範囲 | pp.23-28(NLP), |
| ページ数 | 6 |
| 発行日 | 2020-05-08 (NLP) |