Original link: https://www.cnblogs.com/chamie/p/5579884.html
Basic types of neural networks
The basic types and learning algorithms of neural networks:
At present, there are dozens of neural network models, which can be divided into three categories according to the network structure: feedforward network, feedback network and self-organizing network.
Feedforward neural network refers to the layered arrangement of neurons, which are composed of input layer, middle layer and output layer. The neurons in each layer only accept input from the neurons in the previous layer, and the subsequent layers have no signal feedback to the previous layer. The input pattern is propagated sequentially through each layer, and finally the output is obtained on the output layer. This kind of network structure is usually suitable for prediction, pattern recognition and nonlinear function approximation. Generally, the typical forward neural network is based on the gradient algorithm neural network such as BP network, and the optimal regularization method such as SVM, radial basis neural network and extreme learning Machine neural network.
The feedback network is also called the regression network. The input signal determines the initial state of the feedback system. The system gradually converges to an equilibrium state after a series of state transitions. Therefore, stability is one of the most important indicators of the feedback network. The typical one is the perceptron network. , Hopfield Neural Network, Hamming Neural Network, Wavelet Neural Network, Bidirectional Associated Storage Network (BAM), Boltzmann Machine.
Self-organizing neural network is a learning network without a teacher. It simulates the behavior of the human brain and automatically adapts to unpredictable environmental changes based on past experience. Because there is no teacher signal, this type of network usually adopts the principle of competition for network learning.
Other articles:
https://baijiahao.baidu.com/s?id=1656767454769642971&wfr=spider&for=pc
http://www.360doc.com/content/17/0604/12/43851902_659777911.shtml
https://www.sohu.com/a/227750314_642183
Popular Science | Three categories of deep learning algorithms: CNN, RNN and GAN