Improved artificial bee colony algorithm and its application in data clustering

Artificial Bee Colony (ABC), as a new swarm intelligence based method, suffers from low precision and efficiency in solving optimization problems. Inspired by the improved strategies of Particle Swarm Optimization (PSO), we have proposed some modification on the original ABC iteration equation. In this paper, inertial weight is added on the first item which balances the local and the global searching processes. The contractive parameter is also introduced to the second item instead of the random number, which shows the nonlinear descending characteristic and has contractive effect on the search space of the algorithm. Furthermore, an additional random disturbance item is added to the renewal equation of the basic ABC algorithm, which helps the algorithm continue to search in the later iteration stage and continually increases its accuracy. The new improved ABC (IABC) method is firstly used in benchmark function optimization to test the performance and then it is applied to data clustering analysis of the DNA microarray gene expression data and PPI data sets. The simulation results show that the IABC is more effective than the state-of-the-art methods.