Applied Convex Optimization with SVM

APPLIED CONVEX OPTIMIZATION ASSIGNMENT

Abstract

A Linear Support Vector Machine was created to classify vectors to either one of the two classes. The vectors have two dimensional coordinates making it possible to create clear visual representations of the data. In order to create this SVM, first a convex optimization problem was formulated. The SVM was modeled with a favourite off-the-shelf solver and evaluated. Also an own-made algorithm based on the (sub)gradient descent algorithm has been implemented to solve this convex problem. These two implementations are evaluated on performance in respect to number of iterations, CPU time, and convergence of the algorithms.

Applied_Convex_Optimization_Essay

Download PDF