基于人工免疫机制的RBF网络混合训练算法

基于人工免疫聚类机制和免疫进化算法，提出了一种新型的设计RBF网络的混合算法。该方法利用人工免疫聚类机制，根据输入数据集合自适应地确定RBF网络核函数的数量及其中心的初始位置。采用免疫进化算法训练RBF网络，进一步缩小了标准进化算法搜索空间的范围，提高了算法的收敛速度。计算机仿真表明，这种RBF网络结构精简并具有较强的泛化能力。

Hybrid RBF training algorithm based on artificial immunology

Based on artificial immune clustering and Immune Evolutionary Algorithm (IEA), a novel hybrid RBF design method is proposed. The artificial immune clustering is used to adaptively specify the amount and initial position of centers of basis functions in RBF network according to input data set. Then immune evolutionary algorithm is used to train the RBF network, which reduces the searching space of canonical evolutionary algorithm and improves the convergence speed. Computer simulations demonstrate that the RBF network designed in this method has a concise structure with good generalization ability.