Clustering on the Basis of a Divisive Approach by the Method of Alternative Adaptation

The paper presents a combined approach to clustering based on the integration of the divisive method with the methods of collective alternative adaptation. This group of methods is characterized by the sequential separation of the original cluster consisting of all objects, and the corresponding increase in the number of clusters. The objective is to enhance the convergence of the algorithm and the ability to exit from local optima, which allows you to work with large-scale problems and get high-quality results in a reasonable time. In this paper, the process of finding a solution is represented as an adaptive system. Under the influence of a series of adaptive actions, all objects (collective) are successively redistributed between the clusters. The goal of a specific object m(i) is to reach a state in which the total vector of forces acting on it from all objects placed in the same cluster with mi had the maximum value. The goal of the collective of objects is to achieve such a separation of objects into clusters, at which the minimum distance between a pair of objects belonging to different clusters has a maximum value. To implement the adaptation mechanism, each object mi is assigned an adaptation machine AAi. Studies have shown that the time complexity of the algorithm at one iteration has an estimate of O (n(2)), where n is the number of objects.