摘要
基于事物演化发展的思想,尝试对传统RBF神经网络的结构进行了优化。首先从IPL算法对RBF网络的学习训练不足出发,通过调整RBF神经网络基函数的采样算子,得到一个规模可以控制的网络模型,最后给出了仿真验证结果。
Evolvement is a basic property of things development and consists of a series of complex change activities.Based on this fact,this paper focuses on providing a viable alternative for RBF network.First,aiming at the Incremental Projection Learning （IPL） algorithm＇s disadvantage,the improved IPL algorithm for constructing RBF network based on the adjustment of the sampling operator is presented,and induces a simpler network structure than before.The simulation result demonstrates the effectives of the improved RBF network.
出处
《计算机工程与应用》
CSCD
北大核心
2007年第1期94-95,124,共3页
Computer Engineering and Applications
基金
国家部委预研基金项目。
