进化计算在神经网络中应用

本文介绍国外关于进化计算和人工神经网络结合技术的研究现状,主要包括进化计算用于神经网络优化设计、神经网络输入数据预处理、选择性神经网络集成等方面的内容,其中优化网络设计包括三个层次:权值优化、结构优化和学习规则优化;数据预处理主要用于模式分类中进行特征提取;选择性的网络集成则能很好地提高集成后网络的泛化能力.并对研究过程中出现的主要问题及未来发展趋势进行了讨论.

Survey on the application of evolutionary computation into artificial neural networks

This paper discusses and summarizes the present status of combining evolutionary computation with artificial neural networks, which mainly includes the following three parts: the application of evolutionary computation to the design optimization of neural networks; the preprocessing of input data; and integration of optional neural networks. The first part includes optimizations of weight, structure and learning rules, while the data processing is mostly used to distill the characteristics that occur in mode-sorting. And lastly, the integration of networks would profoundly improve its ability of generalization. In addition, the essay also discusses some issues which occurred during research, and states some opinions about the future of artificial neural network.