Statistical Pattern Recognition DML统计模式识别.pdfVIP

Statistical Pattern Recognition Support Vector Machine (SVM) Hamid R. Rabiee Hadi Asheri, Jafar Muhammadi, Nima Pourdamghani Spring 2012 /courses/90-91/2/ce725-1/ Agenda Introduction Some Concepts Linear, Hard-Margin SVM Linear, Soft-Margin SVM Nonlinear SVM SVMs vs. Neural Networks Conclusion 2 Sharif University of Technology, Department of Computer Engineering, Pattern Recognition Course Support Vector Machines Decision surface is a hyperplane (line in 2D) in feature space. SVM three main ideas: Define what an optimal hyperplane is: maximize margin Extend the above definition for non-linearly separable problems: have a penalty term for misclassifications Map data to high dimensional space where it is easier to classify with linear decision surfaces: reformulate problem so that data is mapped implicitly to this space 3 Sharif University of Technology, Department of Computer Engineering, Pattern Recognition Course Concepts Which of the hyperplanes should we 2 # e r u t choose? a e F Intuitively, a hyperplane that passes too close to the