Topical issue

Model and method for constructing a heterogeneous cluster ensemble

V. B. Berikov<sup>ab

a</sup> Novosibirsk State University, Novosibirsk, 630090 Russia
^b Sobolev Institute of Mathematics, Novosibirsk, 630090 Russia

Abstract: We consider the problem of data clustering using a heterogeneous ensemble with the use of a co-association matrix. A probabilistic model is stated that takes into account the correlation of evaluation functions with the help of which relationships are found between the characteristics of the ensemble and the quality indicators of the final solution. An expression for the optimal weights of basic algorithms for which the upper bound on the clustering error probability estimate is minimal is found. An experimental study of the proposed method has been carried out showing the method to be advantageous over a number of similar methods.

Keywords: cluster analysis, ensemble of algorithms, co-association matrix, cluster ensemble model, optimal weights of algorithms.

Presented by the member of Editorial Board: A. A. Lazarev

Received: 01.02.2022
Revised: 25.06.2022
Accepted: 29.06.2022

DOI: 10.31857/S000523102212008X

 English version: Automation and Remote Control, 2022, 83:12, 1944–1958

Bibliographic databases:

•