Dynamic local search algorithm for the clustering problem

Dynamic clustering problems can be solved by finding several clustering solutions with different number of clusters, and by choosing the one that minimizes a given evaluation function value. This kind of brute force approach is general but not very efficient. We propose a dynamic local search that solves the number and location of the clusters jointly. The algorithm uses a set of basic operations, such as cluster addition, removal and swapping. The clustering is found by the combination of trialand-error approach of local search. The algorithm finds the result 30 times faster than the brute force approach.