Cooperative Micro Artificial Bee Colony Algorithm for Large Scale Global Optimization Problems

Abstract

Large scale optimization problems or optimization problems involving high-dimensions often appear in real world application scenario. The mathematical representation of these problems appears similar to that of traditional low dimensional problems but they exhibit high interdependencies among the variables to be optimized. Hence normal evolutionary algorithms or swarm intelligence based methods cannot be directly operated on these problems to find global optimum. In these situations, cooperating approaches are proved to be very valuable, since they are based on though simple yet power strategy “divide and conquer”. Though handy, computational burden of cooperative approach oriented methods will be high, as they involve optimization of various subcomponents for predefined number of steps. On other hand, recently evolved Micro Evolutionary Algorithms (micro-EAs) are shown to be very powerful strategies for solving optimization problems, as they involve very small population of just a few individuals. This advantage of micro-EA is accompanied by its tendency towards to get stuck in local optima. Hence this paper tries to combine the advantages of both cooperative strategies and also micro-EAs nature accompanied with a swarm intelligent Artificial Bee Colony (ABC) algorithm as main optimizer, to solve optimization problems of very high dimension. The proposed variant is termed as “Cooperative Micro-Artificial Bee Colony” (CMABC) algorithm. Computer simulations over benchmark suite considered and also extensive comparisons over cooperative variants of state-of-art Differential Evolution method show the superiority of proposed algorithm.