An Improved Artificial Bee Colony (ABC) Algorithm for Large Scale Optimization

Artificial bee colony (ABC) algorithm as a new optimization algorithm invented recently has been applied to solve many kinds of combinatorial and numerical function optimization problems. The existing forms of ABC algorithms perform well in most cases. However, ABC algorithm is still lack of capacity for optimizing high dimensional problems without taking the interactions within each dimensional variables into consideration. Inspired by Cooperative Coevolution (CC), this paper adjusts ABC algorithm with cooperative coevolving which we call CCABC. Iteratively, CCABC can discover the relations of the high dimensional variables, considering those relationship dimensions as the same group, and then CCABC optimizes the whole group instead of a single dimension. We test CCABC algorithm on a set of large scale optimization benchmarks and compare the performance with that of original ABC algorithm and two classic CC frameworks CCVIL and DECC-G. Experimental results show that CCABC algorithm outperforms CCVIL, DECC-G, and original ABC algorithm in almost all of the experiments and can solve large scale optimization problems efficiently.