维规约技术综述  

Survey on dimension reduction techniques

作　　者：许明旺[1] 施润身[2] 

机构地区：[1]同济大学电子与信息工程学院,上海200433 [2]同济大学研究生院,上海200092

出　　处：《计算机应用》2006年第10期2401-2404,共4页journal of Computer Applications

摘　　要：从属性选择和维变换两个方面对维规约技术进行了概括。首先对属性选择的基本思想和常用算法进行简要介绍;然后对维变换技术中的几种应用最广泛的方法进行了详细研究,主要包括主成分分析及其相关算法、独立成分分析、因子分析、投影寻踪等方法,简要给出了这些方法间的联系和区别,最后指出了维规约技术的现状和发展方向。Dimension reduction techniques were discussed from the two aspects： feature selection and dimension transformation. Firstly, the basic theory and famous algorithms of feature selection were roughly introduced. Then, several most popular dimension transformation techniques were analyzed in detail including Principal Components Analysis and its related methods, Independent Components Analysis, factor analysis, projection pursuit, etc. Meanwhile, connection and distinction between them were provided. Finally, the present situation and future development of dimension reduction techniques were pointed out.

关 键 词：维规约 主成分分析 独立成分分析 因子分析 投影寻踪 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]