Dynamic Multi-Dimensional Jaguar Algorithm with Adaptive Step for Function Optimization Problem

• Main Authors: Yang, LI-SHENG, 楊立聖
• Other Authors: CHOU, YAO-HSIN
• Format: Others
• Language: en_US
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/7h595f

碩士 === 國立暨南國際大學 === 資訊工程學系 === 107 === Jaguar algorithm (JA) has excellent performance in solving optimization problems. Different from traditional metaheuristic algorithms, JA has great abilities both in exploitation and exploration. Therefore, JA can find the best solution more quickly and efficiently. However, JA has some defects. JA considers only one dimension at a time, therefore it requires more evaluations. Moreover, JA does not adjust its step according to the distance of prey. To solve these problems, this study proposes Multi-dimensional Jaguar Algorithm (MJA) can consider multiple dimensions simultaneously and hunt all preys at the same time. MJA includes the adaptive step, which is adjusted according to the level of the solution. In this way, MJA can adapt its step in every dimension and can rush to its prey more accurately, which allows it to find the local optimal solution more efficiently than traditional JA. Furthermore, MJA uses information from multiple territories to jump diagonally to find the global optimum more accurately and efficiently. MJA, therefore, has stronger search ability than traditional JA. The self-analysis and experiments of this study show that the performance of MJA is better than traditional JA and other metaheuristic algorithms.