基于维数约简与特征选择的PGS方法在数据处理中的应用

PGS Based on Dimension Reduction and Feature Selection for Microarray Data

特征选择和维数约简在机器学习、模式识别和数据挖掘领域是很常用的方法。它们之间也具有一定的联系。但对它们的融合应用1／1前很少研究，从而融合特征选择和维数约简的思路被提出。该思路融合了主成分分析方法和遗传算法，提出PGS方法。并把它应用于基因microarray数据的预测分类，取得了较好的效果。

Feature selection and dimension reduction are two common methods Data Mining,and they have many relations, but there are little attention about how for Machine Leaming,Pattern Recognition and to fusion them, so the method of fusing the leature selection and dimension reduction is proposed.The method fused the Genetic Algorithm and Principal Component Analysis （PGS）, then used it for the classification of gene microarray data.