期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2021
卷号：99
期号：4
语种：English
出版社：Journal of Theoretical and Applied
摘要：General gradient-based optimization techniques such as the steepest descent method, Newton's method, and quasi-Newton method, often fail in globally solving non-convex optimization problems (multi-modal functions). The main reason is that once a local solution has been determined, these methods do not know how to pass a hill to obtain another better local solution. Therefore, a new gradient type method so-called ABCED Steepest Descent Method has been introduced in this paper which does not have the weakness been mentioned above. The ABCED SD method is a hybrid from a modified steepest descent method and the Artificial Bee Colony (ABC) algorithm. ABCED SD method developed through the ABC framework by replacing the exploitation search phase with modified steepest descent method, therefore the global optimum solution is obtained by the modified steepest descent algorithm. Reported numerical results shown that ABCED SD method able to locate the global optimum solution for the benchmarked general global optimization problems and the comparison results with the original Artificial Bee Colony algorithm also shown that ABCED SD method able to obtain the global optimum solution with less iterations. Besides that, ABCED SD method also does not require any initial point and it will not only determine the local minimizer as in the classical steepest descent method, but it manage to determine the global minimizer of the general global optimization problems.
