Indian Institute of Technology Madras

About this Lecture

This lecture covers Support Vector Machines (SVMs) with a focus on the kernel trick, soft margins, and C-regularization. We explore how SVMs map data to higher dimensional feature spaces for linear separability.

Topics include: Linear SVMs, Polynomial & RBF kernels, Slack variables, Hinge loss, Multi-class classification with SVMs, and practical implementation in scikit-learn. Students should have completed Assignment 4 on Logistic Regression.