Skip to content

Latest commit

 

History

History
20 lines (13 loc) · 825 Bytes

readme.md

File metadata and controls

20 lines (13 loc) · 825 Bytes

Kernel SVM

This repository contains the code for a simple kernel-svm that is used to fit a data that looks like sun and mountains.

This work was done as an assignment of the course CS559 by Professor Erdem Koyuncu of University of Illinois, Chicago.

I don't have the original question, but the sun and mountains can be seen in the result image.

Details

Description of the data set randomly chosen based on given target function:

  • In-set accuracy by maximum class classifier: 57%
  • In-set accuracy using SVM classifier: 100%

Number of SVMs computed: 14
Number of SVMs in –ve class: 8
Number of SVMs in +ve class: 6

Kernel Used: Radial Basis Function, gamma = 3

sun and mountains