径向基概率神经网络的一种自组织学习算法

介绍了径向基概率神经网络 (RBPNN)的一种自组织学习算法 ,该算法把径向基概率神经网络的结构原理与自组织聚类算法相结合 ,不仅能够完成对训练样本的聚类分析 ,标识出训练样本的类别属性 ,而且能够自动完成基于该训练样本集的径向基概率神经网络的训练过程 .本算法用于对 IRIS三种花型识别在训练阶段达到 97.33%的识别效果 ,而在推广能力方面 ,由本文算法得到的 RBPNN优于有标识的训练样本的 RBFNN

Self-Organized Learning Algorithm of the Radial Basis Probabilistic Neural Network

The paper introduces a self organized learning algorithm of the radial basis probabilistic neural network (RBPNN). The proposed algorithm, which integrates the structure principle of the RBPNN and the self organized clustering algorithm, not only accomplishes the clustering of the training samples i.e., explicitly labels the class properties of the training samples, but also at the same time automatically completes training of RBPNN. In the application of IRIS classification problem, 97.33% of the recognition rate is achieved in the training stage, furthermore in the generalization ability the RBPNN learned by the proposed algorithm is better than the result by the radial basis function neural network (RBFNN).