一种新的基于PCA的集成学习算法

Algorithm of PCA Based Ensemble Learning

给出了主成分分析法(PCA)的数学描述及解释,提出了基于PCA的分类器提取方法及基于PCA的集成学习算法.在UCI的6个公用数据集上,对提出的算法进行了较全面的实验研究和分析,实验表明在多项指标上所提出的算法优于表现良好的传统集成学习算法.

The mathematical description about the principal components analysis method （PCA） is given. The approaches of PCA based classifiers training and the algorithm of PCA based ensemble learning are put foward. Experimental results on six common data sets of UCI show our algorithm is more efficient than the traditional ensemble learning algorithms.