DOCUMENT CLASSIFICATION VIA TEXTCC BASED ON STEREOGRAPHIC PROJECTION
TextCC can classify real documents instantly by cosine similarity. In this paper, stereographic projection is defined from n dimensional real space to the surface of the unit sphere in (n+1) dimensional space. This paper also proposes the relation between the euclidean distance in n dimensional space and the cosine similarity in (n+1) dimensional real space. To classify documents with represented vectors normalized by stereographic projection, modification on the construction of the weight matrix of hidden layer of TextCC and the fundamental for those modifications are presented. With those modifications, TextCC can classify real documents instantly by Euclidean distance. Experimental results show that TextCC can classify real documents well by Euclidean distance based on stereographic projection.
Stereographic projection cosine similarity TextCC
ZHEN-YA ZHANG SHU-GUANG ZHANG XU-FA WANG
MOE-Microsoft Key Laboratory of Multimedia Computing and Communication, University of Science and Te Statistics&Finance Department of USTC, 230027, Hefei, China Computer Science Department of USTC, 230027, Hefei, China
国际会议
2006 International Conference on Machine Learning and Cybernetics(IEEE第五届机器学习与控制论坛)
大连
英文
1368-1372
2006-08-13（万方平台首次上网日期，不代表论文的发表时间）