Adaptive trajectory analysis of replicator dynamics for data clustering

Abstract

We study the use of replicator dynamics for data clustering and structure identification. We investigate that replicator dynamics, while running, reveals informative transitions that correspond to the significant cuts over data. Occurrence of such transitions is significantly faster than the convergence of replicator dynamics. We exploit this observation to design an efficient clustering algorithm in two steps: (1) Cut Identification, and (2) Cluster Pruning. We propose an appropriate regularization to accelerate the appearance of transitions which leads to an adaptive replicator dynamics. A main computational advantage of this regularization is that the optimal solution of the corresponding objective function can be still computed via performing a replicator dynamics. Our experiments on synthetic and real-world datasets show the effectiveness of our algorithm compared to the alternatives.