3.2 Support Vector Machines — Geometry of Maximum Margin

This video explains the geometric idea behind Support Vector Machines and why the classifier is defined by the maximum margin. By comparing SVM with logistic regression, we show how the margin arises naturally from the optimization problem.

Watch the full series:
https://www.youtube.com/playlist?list=PLdMM2W7mxqIYDURvh37jLz8sd8sDjonk5

0:00 SVM Setup and Margin Intuition
4:27 Logistic Loss vs Hinge Loss
8:29 Where the Margin Comes From
15:49 Optimization and Geometry of SVM

Topics:
• Support Vector Machines (SVM)
• Maximum margin classifier
• Linear classifiers in feature space
• Hinge loss vs logistic loss
• Primal formulation of SVM
• Hard margin SVM
• Soft margin SVM
• Geometric interpretation of margin
• Support vectors
• Margin derivation 2 / |w|
• Constrained optimization in SVM

Видео 3.2 Support Vector Machines — Geometry of Maximum Margin канала Jingyuan Hu