摘要
主成份分析法是用于简化数据的一种技术,现实世界中的数据复杂且庞大,对于某些复杂数据就可应用主成分分析法对其进行简化。文中着重介绍了健壮性KPCA算法并引入了粒度的思想,健壮性KPCA算法能推导出特征空间中信号重组的最小错误标准,并自动识别训练样本集中的无关数据,且经过计算消除它们对KPCA算法准确度的影响。可以将其应用于股票数据中,并将所得的主分量图与原图比较,发现效果明显,由此可看出KPCA算法是一种相当有用的算法。
Principal component analysis(PCA) is a technique used to reduce the complication of data .In the world data is huge and complicated .To some complicated data can use PCA to reduce them.The paper places the emphases on the robust KPCA algorithm. It shows that robust KPCA algorithm can generalize the minimum error criteria of signal reconstruction to feature space and recognize the outliers in the training sample set automatically,then exterminate their effects to the accuracy of the KPCA. In the experiment, use KPCA to analyse data of stock and compare original chart with the result of the experiment.The effect is obvious. So we can know KPCA is a considerably available algorithm.
出处
《微机发展》
2004年第12期129-131,共3页
Microcomputer Development
基金
安徽省教育厅自然科学基金资助项目(2003KJ007)
