國立臺灣大學資訊工程學系Kao, Wei-ChunWei-ChunKaoChung, Kai-MinKai-MinChungSun, Chia-LiangChia-LiangSunLin, Chih-JenChih-JenLin2006-09-272018-07-052006-09-272018-07-052004http://ntur.lib.ntu.edu.tw//handle/246246/20060927122900179258In this paper, we show that decomposition methods with alpha seeding are extremely useful for solving a sequence of linear SVMs with more data than attributes. This strategy is motivated from (Keerthi and Lin 2003) which proved that for an SVM with data not linearly separable, after C is large enough, the dual solutions are at the same face. We explain why a direct use of decomposition methods for linear SVMs is sometimes very slow and then analyze why alpha seeding is much more effective for linear than nonlinear SVMs. We also conduct comparisons with other methods which are efficient for linear SVMs, and demonstrate the effectiveness of alpha seeding techniques for helping the model selection.application/pdf219319 bytesapplication/pdfzh-TWotherhttp://ntur.lib.ntu.edu.tw/bitstream/246246/20060927122900179258/1/linear.pdf