Modified stages of algorithms for computing estimates in the space of informative features

Abstract

In the article the tasks of development of the recognition algorithm steps areformulated, based on calculation of estimates using the Fisher's informative criterion in the space of informative features. The problem of optimization for the selection of informative feature sets using the Fisher-type informative criterion in the training sample having objects belonging to only one class is constructed, the proximity function in the space of informative features is constructed. In determining the values of the proximity function, the elements and properties of the Fisher functional are used, the upper and lower boundaries of the proximity function are also defined and the lemmas, properties and assumptions that support them are formulated.