Speed UP Information Gain Based Online SVM for SPAM Filtering
Guanglu Sun, Yuewu Shen, Haoliang Qi · ASME Press eBooks · 2011
In this paper, we propose IGFS (information gain based feature selection) method to speed up online SVM filter for spam filtering. Though online SVM classifier gives high classification performance on online spam filtering on large benchmark data sets, its computational cost turns out to be very expensive for other faster methods such as Na1ve Bayes on large-scale application. We use information gain based feature selection method to reduce feature vector dimension of online SVM filter in this paper. Experimental results illustrate that the method improves the filter performance a little and greatly reduces the computational cost of online SVM filter.