Опис навчальної дисципліни

Summary of the subject:

The following concepts are considered: classification of basic neural architectures by type of learning algorithms and connection architecture, mathematical models of a neuron, types of activation functions and their properties, selection of activation functions for various recognition and classification tasks, supervised learning, training pairs, principles of training pair selection and their influence on the learning process, capabilities of single-layer perceptrons, capabilities of multilayer perceptrons for data analysis, selection of activation function, two-layer perceptrons, the problem of the negating OR function (XOR), overcoming the As a result of studying this course, the student should know:

Advantages and problems of using neural networks in information processing;
Features of using a neural network approach to solve various information processing problems

be able to:

create emulators and analyze the work of simple neural networks;
have skills in operating programs for emulating neural network structures for information processing.