Numerical_Optimization

The purpose of the project is to design a Support Vector Machine (SVMs) for solution of classification problems.This report mainly talks about the whole structure of SVM including its methodology, three different optimizationmethods (Quadratic Penalty, Augmented Lagrangian and InteriorPoint Barrier) with four different Kernels (linear,polynomial, RBF and Sigmoid). I also talk about how to deal with multi-class classification problem by using themethod called one against all which is widely used in machine learning field. Thea data set I used is Fisher Iris datacontaining three classes, four features and 150 samples. I have manually splited 80% of the data as the traning setand 20% of the data as the testing set.