A HOMOLOGY COBOUNDARY LEARNING ALGORITHM FOR DATA REDUCTION

摘要：

Dimension reduction and feature selection are the crucial methods to deal with intricate data. For highdimensional data, this paper gives a data reduction method to reduce data dimension, and a feature selection method to improve classification precision. The current approaches include PGS method based on dimensional reduction and feature selection, nonlinear dimension reduction method in manifold learning, and an efficient dimension reduction method for times-series data sets, etc. PGS method combines dimension reduction with feature selection; nonlinear dimension reduction method in manifold learning has the ability to catch the locality in data analyzing; dimension reduction method for times-series data sets can use either Euclidean distance or edit distance as the similarity criterion. The homology theory was introduced into this paper.We proposed a homology coboundary learning algorithm for data reduction using the idea of the homology coboundary , then used it for the classification of crystalloid.

关键词： data reduction coboundary homology

作者: Min Xian Fan-zhang Li

作者单位: School of Computer Science & Technology,Soochow University,Suzhou 215006,P.R.China

会议类型: 国际会议

会议名称: 2008高等智能国际会议(2008 International Conference on Advanced Intelligence)

会议地点: 北京

会议语种:英文

在线出版日期: 2008-10-18（万方平台首次上网日期，不代表论文的发表时间）