A cooperative multi-population approach to clustering temporal data

In temporal environments, population-based data clustering algorithms suffer when changes in the data occur during the clustering process. Diversity of the population is lost and memory of the individuals of the population is outdated, making the clusters found before the change non-optimal. This paper proposes a new particle swarm optimisation alternative to clustering temporal data. It combines the dynamic properties of the multi-swarm particle swarm optimisation algorithm with the multi-objective properties of the cooperative particle swarm optimisation algorithm. The proposed alternative is compared to various existing data clustering algorithms which are shortly described in the paper and the results are discussed, including a comparison of four performance measures relevant to the clustering of data.