Tmproved Artificial Bee Colony Algorithm Based On Local Search And Binary System

Artificial bee colony algorithm (ABC) is the latest swarm intelligence based op-timization method, which is superior to other population-based intelligent algorithms. However,the updating formula of artificial bee colony is not good at exploitation. This paper is devoted to enhancing or adding a local search mechanism in two versions of artificial bee colony algorithms. Therefore, we proposed two modified artificial bee colony algorithms.The first proposed algorithm is an improved Hooke-Jeeves artificial bee colony algorithm (IHABC). This new algorithm adjusts initial point selection in Hooke-Jeeves search phrase to enhance the quality of whole population by means of improv-ing upper-middle individuals. The search formulas of employed bees and onlooker bees are also altered in order to increase exploitation capacity, In addition, the whole algorithm adopts a new objective function evaluation standard, which works well on any function. We choose 30 functions from benchmark function library to test the effectiveness of our proposed algorithm IHABC. Compared with artificial bee colony algorithm and Hooke-Jeeves-based artificial bee colony algorithm, the obtained re-sults clearly indicate that IHABC gets the higher approximate solution precision in solving unconstrained optimization problems.The second proposed algorithm is an improved binary artificial bee colony algo-rithm (IGB-ABC). On the basis of binary artificial bee colony algorithm (GB-ABC), we use maximum-minimum distance method to initialize so as to make initial points distributed uniformly. We adopt a variety of candidate solution updating formulas and add an extra local search phrase. Five clustering images and five data sets are used to test the effectiveness of our proposed algorithm IGB-ABC. The obtained results clearly indicate that IGB-ABC is much better than GB-ABC algorithm in clustering validity index, dynamic clustering number and clustering results.