ZhengZerong / COTAProject

Project for the course "Convex Optimization": A Toy Example of SVM-GSU Implemented in Matlab

Final Project for the Course COTA: a toy example of SVM-GSU

A toy example of SVM-GSU, which is described in Linear Maximum Margin Classifierfor Learning from Uncertain Data, TPAMI 2018.

Requirement: Matlab + CVX

Usage: run main.m in Matlab console.

Illustration

Illustration on 2D data.

• Red/green points: two classes of data points.
• 1st image: comparison of SVM-GSU (pink solid line) with standard SVM (blue dash line).
• 2nd/3rd image: comparison of SVM-GSU with standard SVM (black dash line) that learns by sampling 100/1000 points from the input Gaussian.

Related Link

• The paper of SVM-GSU: link
• A full C++ implementation of SVM-GSU provided by the authors: link