|
|
artificial neural network (ANN).
Ключевые слова: искусственный интеллект, искусственная нейронная сеть, распознавание объектов, алгоритм, обучение.
Keywords:artificial intelligence, artificial neural network, Pattern Recognition, algorithm, training.
In the theory of artificial intelligence, in the quest to simulate the ability of biological nervous systems formed the idea of neural network (NN). Currently, research in this area are mainly used inapplications where formal analysis would be difficult or impossible to use. Thesetechnologies based on the principle of machine learning system which is used in various disciplines of cognitive science, forecasting, statistics, and management areas, nonlinear identification, robotic systems, diagnostics in medicine and engineering, pattern recognition, etc.
An artificial neural network is a circuit composed of many simple processing units referred to as neurons, which are inspired by their (much more complex) biological counterparts in the brain. Each neuron receives inputs from several other neurons; it then performs some basic processing on these inputs and passes the result on to other neurons. Each connection typically has a weight associated with it, which affects the strength of the signal that passes along this connection. With the right choice of neurons, connections, and weights, a neural network can do any processing you like, but how would you come up with the right choice? Neural networks are so popular because researchers have devised ways of automatically learning certain aspects of network design, particularly the value of the weights, just by showing examples of the problem to the network.
Fig. 1.A neural network for recognizing 3 different alphabetic letters.
227
|
