Neural Network Model
The neural network is constituted by a large number of nerve element interconnect network. According to the interconnection of the neurons in thenetwork, the common network structure can be divided into the following three categories:
1.Feedforward Neural Networks
Feedforward network, also known as the forward network. Such a network there will be only in the training process the feedback signal transmission in theclassification process, the data can only move forward, until you reach the outputlayer, the layer did not back the feedback signal is called a feedforward network.Perceptron (perceptron) with BP neural network is a feedforward network.Figure 1 is a three-layer feedforward neural network, where the first layer is theinput unit, the second layer called the hidden layer, the third layer is called theoutput layer..
For a three-layer feedforward neural network N, with X the input vector of thenetwork, said ~ the W3, W1, the connection weight vector of the network layers, F1~ the F3 layer neural network activation function.
The first layer of neurons in the output of the neural network as follows:
O1 = F1( XW1 )
The output of the second layer:
O2 = F2 ( F1( XW1 ) W2 )
The output of the output layer:
O3 = F3( F2 ( F1( XW1 ) W2 ) W3 )
2. Feedback Neural Networks
Feedback neural network is a neural network with feedback connections fromoutput to input, and its structure is much more complex than the feedforwardnetwork. Typical of the feedback neural network: Elman network and the Hopfieldnetwork.
3. SOM ,Self-Organizing Neural Networks
Self-organizing neural network is an unsupervised learning network. It does this byautomatically find a sample of the inherent law and the essential attribute of self-organization, adaptive to change the network parameters and structure.
