Research on Parallel Optimization of Artificial Bee Colony Algorithm

In order to reduce running time of traditional artificial bee colony algorithm when the population size is large, parallel processing is done on CPU and GPU respectively in this paper. Parallel computing is used in computing the value of initial fitness and employed bees phase, and the consuming time of each iteration through parallel algorithm proposed in this paper are observed. Under the different populations, the performances of the three algorithms are evaluated, and the speedup ratios are compared. The results show that the parallel artificial bee colony algorithm can greatly reduce the running time by this method.