Improving K-Means with Harris Hawks Optimization Algorithm

Data clustering aims at partitioning the data set into several disjoint segments, where the data inside the same segment are closely related, and the ones between two different segments are not. K-means is a popular data clustering algorithm, but it is easy to fall into the local optimum. To improve K-means' performance, in this work, a Harris Hawks Optimization algorithm (HHO), which is a newly emerging Evolutionary Algorithm with strong global search capability, is utilized to optimize the central data set. In particular, we encode all the central data in one individual, and optimize them with HHO. In the experiment, multiple data sets are utilized to test our proposal's performance. The experimental results show that our approach significantly outperforms the traditional K-means.