Department of Computer Science
On solving complex optimization problems with objective decomposition
This paper addresses the complex optimization problem, of which the objective function consists of two parts: One part is differentiable and the other part is non-differentiable. Accordingly, we decompose the original objective function into several relatively simple sub-objective ones, which subsequently formulate as a multiobjective optimization problem (MOP). To solve this MOP, we propose a simulated water-stream algorithm (SWA) inspired by the natural phenomenon of water streams. The water streams with a hybrid process of downstream and penetration towards the basin is analogous to the process of finding the minimum solution in an optimization problem. The SWA featuring a combination of deterministic search and heuristic search generally converges much faster than the existing counterparts with a considerable accuracy enhancement. Experimental results show the efficacy of the proposed algorithm. © 2013 IEEE.
Multi modal, Non-differentiable function, Objective decomposition, Simulated waterstream algorithm
Source Publication Title
Proceedings of 2013 IEEE International Conference on Systems, Man, and Cybernetics SMC 2013
Manchester, United Kingdom
Link to Publisher's Edition
Cheung, Yiu-Ming, and Fangqing Gu. "On solving complex optimization problems with objective decomposition." Proceedings of 2013 IEEE International Conference on Systems, Man, and Cybernetics SMC 2013 (2013): 2264-2269.