基因算法辅助人工神经元网络的结构设计

Genetic algorithm aided architecture design of artificial neural network

提出了以基因算法为基础的人工神经元网络结构设计基础。首先从连接权的演化入手 ,研究了基因算法的实现过程。在结构设计中 ,提出了稀疏化的编码方法。仿真结果表明这种优化方法对于神经网络的选取是有效的。

A system for the architecture design of artificial neural network based on the genetic algorithm(GA) is proposed. With the evolution of connection weights first, the author researches into the materialization of GA, and a combined genetic/back propagation learning algorithm is hence proposed. In the development of neural networks architectures, a sparse coding is put forward. Simulation results show that this optimization method is efficient for selecting ANNs structures.