Improving Model Adaptation for Permutation Problems by RTR with Model-associated Method
| Autor: | Jia-Hui Wu, 吳家輝 |
|---|---|
| Rok vydání: | 2015 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 103 This thesis integrates the restricted tournament replacement (RTR) into estimation of distribution algorithms (EDAs) to effectively solve permutation problems and provides a method for choosing the distance measure of RTR. Since the semantics of permutation problems differ, the performance depends on distance-measure choosing. Inspired by RTR and the technique of the model adaptation for EDAs on permutation problems, the proposed method chooses the distance measure according to the selected model. We also investigate the setting of window size and modify reward policy by the feedback from our proposed method. Some experiments are used to test the efficiency of the proposed method. The results indicate that EDAs for permutation problems with RTR using the correct distance measure outperform the originals, semantic niching technique provides additional knowledge to learn the semantics of permutation problems and distance-measure choosing helps to model adaptation. Combining these features, our proposed method is close to the best performing model and may outperform any single models on problems without semantic model specially designed for that. In this thesis, our proposed method improves the performance by 16 percent in average. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|