01. Margin, Hard Margin Linear SVM

Support Vector Machine (SVM)

Separating Hyperplane

Geometric Margin

Convex Optimization Problem

Lagrangian Formulation

Lagrangian Dual

Characteristic of the Solutions

Classifying New Data Points

02. Soft Margin SVM, Nonlinear SVM, Kernel

Linearly Nonseperable Problems

Convex Optimization Formulation

Soft Margin SVM Classifiers

Lagrangian Formulation

Lagrangian Dual

Characteristics of the Solution

Soft Margin SVM Classifiers

Kernel Methods for Nonlinear Classification

Nonlinear Decision Boundary

Transforming Data

Transforming Data - Exmaple

Mapping Original Space to Kernel Space

Kernel Mapping

Kernel Mapping - Example

Kernel Functions

Example of Nonlinear SVM Using Kernel Function

Choosing Kernel Functions

Nonlinear(Kernel) SVM Classifiers

Linear vs Nonlinear SVM Classifiers

[핵심 머신러닝] SVM 모델 1 (Margin, Hard Margin Linear SVM)
[핵심 머신러닝] SVM 모델 2 (Soft Margin SVM, Nonlinear SVM, Kernel) by 김성범 교수 강의자료 참고