SVM-Kernels

Goal of this project is to implement perceptron,Dual perceptron,Linear Kernel and RBF kernel without using any Machine Learning Libraries

Perceptron and Dual Perceptron

1. I Implemented the perceptron algorithm and dual perceptron algorithm on PerceptronData dataset and ran ten fold cross-validation and then compared the performance of the two algorithms on the PerceptronData dataset and made sure that they have (almost) identical performance.

Kernelizing Dual Perceptron

1. I implemented dual perceptron with the linear kernel on the Two Spiral dataset and showed that the data is not separable using ten-fold cross validation.

Radial Basis Function Kernel (RBF kernel)/ Gaussian kernel

1. I implemented dual perceptron with RBF kernel on the Two Spiral dataset and showed that the data is separable using ten-fold cross validation.

About dataset:

• PerceptronData: This is a binary classification dataset consisting of four features and the classes are linearly separable. • Two Spirals: This dataset has two features and it is non-linearly separable