A Moment Bound for Multi-Hinge Classifiers
Abstract
The success of support vector machines in binary classification relies on the fact that hinge loss employed in the risk minimization targets the Bayes rule. Recent research explores some extensions of this large margin based method to the multicategory case. We show a moment bound for the so-called multi-hinge loss minimizers based on two kinds of complexity constraints: entropy with bracketing and empirical entropy. Obtaining such a result based on the latter is harder than finding one based on the former. We obtain fast rates of convergence that adapt to the unknown margin.
Cite
Text
Tarigan and van de Geer. "A Moment Bound for Multi-Hinge Classifiers." Journal of Machine Learning Research, 2008.Markdown
[Tarigan and van de Geer. "A Moment Bound for Multi-Hinge Classifiers." Journal of Machine Learning Research, 2008.](https://mlanthology.org/jmlr/2008/tarigan2008jmlr-moment/)BibTeX
@article{tarigan2008jmlr-moment,
title = {{A Moment Bound for Multi-Hinge Classifiers}},
author = {Tarigan, Bernadetta and van de Geer, Sara A.},
journal = {Journal of Machine Learning Research},
year = {2008},
pages = {2171-2185},
volume = {9},
url = {https://mlanthology.org/jmlr/2008/tarigan2008jmlr-moment/}
}