Feature shaping for linear SVM classifiers

: ? Feature Shaping for Linear SVM Classifiers George Forman, Martin Scholz, Shyamsundar Rajaram HP Laboratories HPL-2009-31R1 text classification machine learning, feature weighting, feature scaling, SVM Linear classifiers have been shown to be effective for many discrimination tasks. Irrespective of the learning algorithm itself, the final classifier has a weight to multiply by each feature. This suggests that ideally each input feature should be linearly correlated with the target variable (or anti-correlated), whereas raw features may be highly non-linear. In this paper, we attempt to re-shape each input feature so that it is appropriate to use with a linear weight and to scale the different features in proportion to their predictive value. We demonstrate that this pre-processing is beneficial for linear SVM classifiers on a large benchmark of text classification tasks as well as UCI datasets. External Posting Date: May 6, 2009 [Fulltext] Approved for External Publication Internal ...

• Type: Conference
• Year: 2009
• Where: KDD
• Authors: George Forman, Martin Scholz, Shyamsundar Rajaram
• Keywords: 15th Acm Sigkdd, Data Mining, KDD 2009, Linear Svm Classifiers, SVM Linear Classifiers