STOCHASTIC MODELING OF INTEGRATED CLASSIFICATION PROCEDURES

The paper considers the problem of defining the stochastic model of an integrated procedure for object classification, which arises when an odd number of basic classification procedures are used together. To resolve the uncertainty arising from a high inconsistency of the results of individual basic procedures, two different approaches have been proposed. The first approach is to reuse basic procedures until an unambiguous classification result is obtained. The second approach does not require repeated application of basic procedures and is based on resolving uncertainty using the average principle. The application of generating functions of a special type for calculating the elements of a probabilistic matrix of an integrated procedure formed by an arbitrary odd number of basic procedures is considered. A system of decision rules is given that allows to determine the result of an integrated classification using the joint application of the majority principle and the average principle. These approaches can serve as a basis for the software implementation of tools for solving applied problems of object classification using modern data mining methods.

Authors: E. A. Burkov, P. I. Paderno, Е. A. Tolkacheva

Direction: Informatics, Computer Technologies And Control

Keywords: Probability matrix, integrated classification of objects, average principle, majority principle, classification procedure, stochastic model

---

View full article