The term neural networks is used to describe a number of different models intended to imitate some of the functions of the human brain, using certain of its basic structures. The authors aim to convey an intuitive and practical understanding of the topic and to provide the foundations necessary before undertaking further study. To this end, the first part of the book is devoted to a description of biological foundations. Biology is the source of study of neural networks and it seems probable that it will continue to provide a source of essential ideas. Following this introduction, a general model for neural networks is presented and a number of today's most important models are studied. Lastly, a number of real applications are discussed.
Multilayer Neural Networks
7 autres sections non affichées
activation Adaptive algorithm allows application architecture areas associative back-propagation brain calculated called carried cells characteristics characters classification completely connection weights connections Consider consists constructed containing convergence corresponding decision defined described desired developed difficulty distribution enable energy error example external Finally function generalisation given gives hidden layers Hopfield network implement initial input vector interesting internal introduced Kohonen layer learning limited linear machine manner matrix mechanism memory method minimising mode modified modulation necessary neural networks neurons noise obtained operation orthogonal output parallel particular pattern perceptron performance phase possible potential presented probability problem proposed recognition representation represented response rule shown in figure shows signal similar simple simulated solution stable stage step structure term threshold towns units variable visual weight vectors weights