Integrating nature-inspired optimization algorithms to K-means clustering

Abstract

Although K-means clustering algorithm is simple and popular, it has a fundamental drawback of falling into local optima that depend on the randomly generated initial centroid values. Optimization algorithms are well known for their ability to guide iterative computation in searching for global optima. They also speed up the clustering process by achieving early convergence. Contemporary optimization algorithms inspired by biology, including the Wolf, Firefly, Cuckoo, Bat and Ant algorithms, simulate swarm behavior in which peers are attracted while steering towards a global objective. It is found that these bio-inspired algorithms have their own virtues and could be logically integrated into K-means clustering to avoid local optima during iteration to convergence. In this paper, the constructs of the integration of bio-inspired optimization methods into K-means clustering are presented. The extended versions of clustering algorithms integrated with bio-inspired optimization methods produce improved results. Experiments are conducted to validate the benefits of the proposed approach.