A Study on Random Search Method with High Efficiency.
Masahide Nakamura, Akihiro ORISAKA · TRANSACTIONS OF THE JAPAN SOCIETY OF MECHANICAL ENGINEERS Series C · 1999
A new search method to solve the global optimization of continuous variables is proposed. In this method, the tabu search with two separated trajectories was adopted to avoid the trapping in local optimum. Additionally, we utilized the information of gradient to decrease the CPU time. The efficiency of this method was tested by using the ten dimensional problems. Moreover, this method was applied to learning of neural network and it was shown that this method was suitable for the learning algorithm of neural network.