ABSTRACT
Clustering is considered a model of an object in a “fuzzy” language. Sorting of clusters with the aim of finding an optimal cluster is called “self-organization clustering.” Cluster analysis is usually viewed as a theory of pattern recognition "without teacher" i.e., without indication of a target function. The main difference between self-organization modeling and self-organization clustering is the degree of detail of the mathematical language. Clusterization algorithms differ according to their learning techniques that are categorized as learning “without teacher” and learning “with teacher.” The view of clusterization as a model allows us to transfer the basic concepts and procedures of self-organization modeling theory into the self-organization theory of clusterization. As in self-organization modeling, the model with optimal complexity does not coincide with the expert’s opinions. The best cluster, being consistent and optimal according to the regularity precision, does not coincide with a priori specified expert decisions.
