Feature Space Restructuring for SVMs with Application to Text Categorization
Hiroya Takamura, Yūji Matsumoto · 2001
In this paper, we propose a new method of text categorization based on feature space restruc-turing for SVMs. In our method, independent components of document vectors are extracted using ICA and concatenated with the original vectors. This restructuring makes it possible for SVMs to focus on the latent semantic space without losing information given by the original feature space. Using this method, we achieved high performance in text categorization both with small number and large numbers of labeled data. 1