Uniqueness of the SVM Solution
Abstract
We give necessary and sufficient conditions for uniqueness of the support vector solution for the problems of pattern recognition and regression estimation, for a general class of cost functions. We show that if the solution is not unique, all support vectors are necessarily at bound, and we give some simple examples of non-unique solu(cid:173) tions. We note that uniqueness of the primal (dual) solution does not necessarily imply uniqueness of the dual (primal) solution. We show how to compute the threshold b when the solution is unique, but when all support vectors are at bound, in which case the usual method for determining b does not work.
Cite
Text
Burges and Crisp. "Uniqueness of the SVM Solution." Neural Information Processing Systems, 1999.Markdown
[Burges and Crisp. "Uniqueness of the SVM Solution." Neural Information Processing Systems, 1999.](https://mlanthology.org/neurips/1999/burges1999neurips-uniqueness/)BibTeX
@inproceedings{burges1999neurips-uniqueness,
title = {{Uniqueness of the SVM Solution}},
author = {Burges, Christopher J. C. and Crisp, David J.},
booktitle = {Neural Information Processing Systems},
year = {1999},
pages = {223-229},
url = {https://mlanthology.org/neurips/1999/burges1999neurips-uniqueness/}
}