SMpeaks: a semi-supervised clustering algorithm based on density peaks

Clustering by fast search and find of Density Peaks (referred to as DP) was introduced by Alex Rodriguez and Alessandro Laio. DP algorithm is based on the idea that cluster centers are characterized by a higher density than their neighbors and by a relatively large distance from points with higher densities. This algorithm can discover clusters regardless of their shapes and the dimensions of the space containing them. However, it cannot effectively detect clusters with different sizes and densities of arbitrary shapes, especially the same cluster with multiple peaks. Moreover, the DP algorithm needs to select the centers of the clusters by using a decision graph manually. Despite a highly improved performance in semi-supervised clustering, to address this problem, we propose a semi-supervised framework for DP, namely SMpeaks, by integrating pairwise must-link and cannot-link constraints to guide the clustering procedure. We tested the SMpeaks algorithm on complex data sets having clusters with arbitrary shapes, different sizes, and densities. The experimental results have demonstrated that this algorithm is more effective in finding clusters of complex shapes and different densities than DP.