Chetyrbok P. Recognitions of objects on the basis of vectorial measure of closeness of patterns in error space

In this work the methods of neuron networks theory have been grounded and developed concerning the solution of the task of objects recognition with the use of criterion on the basis of vectorial measure of closeness of patterns in errors space. 1. The decision rule has been built for classification of objects in the form of electro-optical images, which enables the reliable recognition under the conditions of presence of noise. 2. The application of criterion on the basis of vectorial measure of closeness of patterns in space of errors allowed to shorten the quantity of classification errors of objects of electro-optical images. 3. The result of the work are applied in the educational process of the university for preparation of specialists in industry of information technologies.