A Hybrid Ant-Based Clustering Algorithm
Marianne Chong, Mylini Munusamy · 2006
This research examines the ant-based clustering method as an alternative to k-means algorithm. Of particular interest is the AntClust which is modeled after the nestmate recognition system of real ants. We propose an algorithm called the Hybrid Ant-based Clustering Algorithm (HACA) which is a hybrid approach for unsupervised clustering problems. This algorithm combines the features of AntClust and k-means. HACA employs three different strategies — the blacklist strategy, the sniffing strategy and the nests fusion method, to improve the performance of the algorithm. We have conducted experimental investigations to demonstrate the effectiveness of HACA and the results have shown the advantages of the three strategies proposed as well as the improved performance of HACA compared to two other algorithms.