核聚类快速后向传播算法

后向传播神经网络算法是一种经典的分类算法，但是通常该算法训练时间较长。针对这种不足，提出了一种基于核聚类的快速后向传播算法。利用核聚类将原始样本划分为多个簇，对每一个簇计算簇中心样本，利用所有的簇中心样本作为新训练集进行神经网络学习。在UCI标准数据集和说话人识别数据集上的仿真实验，充分说明了算法较传统后向传播算法具有明显的速度优势。

Fast kernel clustering back propagation algorithm

The back propagation algorithm is a classic classification algorithm, but it is usually with a long training time. For this deficiency, this paper presents a fast back propagation algorithm based on kernel clustering. The algorithm uses kernel clustering to divide the original samples into multiple clusters, then computes the sample’s center of each cluster, and uses all the center samples as the new training set and trains a neural network classifier. Simulation experiments on UCI standard data set and speaker recognition data set show that the proposed algorithm has obvious advantages compared with the traditional back propagation algorithm.