Hybrid Bacterial Foraging Algorithm for Data Clustering

Data clustering is a fundamental but challenging task in many research fields, such as pattern recognition, data statistics and machine learning. Various clustering techniques such as K-means have been proposed for compact clustering solutions. However, these clustering methods are sensitive to the selection of initial cluster centers and suffer from the problem of premature convergence. Recently, novel swarm intelligence algorithms such as bacterial foraging algorithm (BFA) offer some inspirations for the design of clustering algorithms, owing to their promising global and parallel search capacities. In this paper, we propose a hybrid clustering algorithm (BFCA) based on BFA and K-means. The proposed algorithm attempts to take full advantages of global search capacities of BFA and excellent local search abilities of K-means. Experimental results on four datasets show that the proposed technique outperforms K-means in terms of clustering quality.