一种全局收敛的PCA神经网络学习算法

主元分析(PCA)也称为K-L变换是进行特征提取的一种重要方法。近年来，为了处理海量数据，许多基于Hebbian学习算法的PCA神经网络被提出来。传统的算法，通常不能保证其收敛性或者收敛速度较慢。基于CRLS神经网络，本文提出了一种新的确保权向量收敛的学习算法，本算法无须在计算中规格化权向量。同时也证明了该学习算法使得权向量收敛到最大特征值所对应的特征向量。实验表明，与传统的CRLS神经网络比较，本文算法准确性得到极大提高。

A Convergent Algorithm for PCA Neural Network

Principal component analysis (PCA)is one of the most general-purpose feature extraction methods. For processing the huge data sets, a variety of learning algorithm for PCA has been proposed. However, traditional algorithms will either divergence or convergence very slowly. Based on the CRLS neural network,a novel convergence algorithm is proposed and the fact that the weight vector will converge to the largest eigenvector is also proved. Finally ,simulation results are also included to illustrate the accuracy of this new algorithm.