Hybrid Optimization Method of Evolutionary Parallel Gradient Search
Zhao Zhongyu, Wenfang Xie, Henry Hong · 2010
In this paper, a global optimization algorithm called Evolutionary Parallel Gradient Search (EPGS) is proposed to search for the global optimum of nonlinear functions. By using the gradient information in Evolutionary Algorithm (EA), EPGS combines two search (optimization) methods in an innovative way such that EA is used to keep the best searches at every step in the optimization process and the gradient descent method is used to update these best searches. EPGS is then tested for its validity on the optimization of benchmark functions and compared to two other hybrid methods (GA with Quasi-Newton and GA with simplex). The results show EPGS has the best performance among the three algorithms.