Synthesis of feature spaces ensuring the quality and reliability of recognition

Solution to the problem of the synthesis of feature spaces that ensure the quality and reliability of recognition is discussed in this paper. The methods of feature selection is inherent for each class of initial properties of the objects, the maximum allowable dimension of the feature space are found, taking into account the size of the training sample, the required quality values and recognition reliability. Based on the proposed methods, an algorithm and software have been developed. Computational experiments on a computer are given and the results are presented in the form of decision rules used for object recognition. The general findings of the study are also given.