A Homogenous Associative Memory Model Based on Structure Learning and Iterative Self-Mapping
基于结构学习和迭代自映射的自联想记忆模型

The connections of traditional ANN (artificial neural network) are fixed, and the learning algorithm is to change the weight of every connection. In this paper, a new model is presented which is based on the characteristic of physiological neuron, and as a unit, a homogenous associative memory neural network was built, the learning algorithm was performed to change the structure of the neural network. The characteristic of this algorithm sets the input and output field of a neuron, adjusts the connection between synapse and axon and parallel iterative self-mapping.The matrix model of the network and the experiment results are also presented in this paper.