Fast evolutionary solution finding for optimization using opposite gradient movement

In this paper, a hybrid algorithm of gradient movement is proposed. On a surface of continuous function, every random point has a gradient value of the function that minimize and convergence to zero when it is a neighborhood with the optimum solution. Each iteration calculates the gradient of function at every point and chooses a minimum gradient point with a shortest distance from the optimum solution to find a new closer candidate to be an optimum point. The comparative experiments were made between CA_PSO, PSO, CACO, and SGA. Results show the proposed algorithm with gradient movement techniques outperforms other.