Neural networks as systems for recognizing patterns (Q1270539)

The main types of neural network models are described in Section 2. Section 3 is devoted to studying the computational potentialities of neural networks, and Section 4 is devoted to the problem of the complexity of learning and calculations in networks. Section 5 describes an elementary type of nonrecurrent network, namely, a perceptron. In other words, we consider a single McCulloch-Pitts neuron with \(n\) inputs. In addition, we analyze the most widely used recurrent model, the Hopfield neural network (a network with a complete set of symmetric synapses between different neurons without self-coupling of neurons). This network is trained with the use of the correlation method of outer product. We use the perceptron and Hopfield neural network to illustrate all the concepts considered. Section 6 provides an analysis of the dynamics of neural networks. Section 7 describes other widespread neural network models, and Section 8 describes neural network generalizations. At the end of each section, we specify the main unsolved problems and the main trends to modern investigations.