PROBABILITY-RESOURCE MODEL OF OBJECT CLASSIFICATION

The paper considers the expansion and use of the stochastic model of object classification for the case when restrictions are imposed on the resource consumption required to perform the classification problem. The principles of formalization of criteria of reliability and validity of procedures and results of classification are briefly stated. A general statement of the classification problem is made on an ordered set of classes using a set of calibrated procedures. A two-stage classification of objects according to the scheme of separate application of procedures is considered and it is shown that, according to the probabilistic criteria of reliability and validity of results, this scheme is identical to the scheme of joint application of procedures, but surpasses the latter in terms of the efficiency of resource consumption. The problem of optimizing the resource consumption for classifying objects is formalized, as well as the problem of optimizing income in the presence of a gain from error-free classification. The proposed probabilistic resource model of object classification allows predicting the classification result and managing costs when using selective sets of procedures.

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

Direction: Informatics, Computer Technologies And Control

Keywords: Probability matrix, two-stage classification, classification procedure, stochastic model, resource management

---

View full article