Stable Sample Compression Schemes: New Applications and an Optimal SVM Margin Bound
Machine Learning
2020-11-10 v1 Machine Learning
Abstract
We analyze a family of supervised learning algorithms based on sample compression schemes that are stable, in the sense that removing points from the training set which were not selected for the compression set does not alter the resulting classifier. We use this technique to derive a variety of novel or improved data-dependent generalization bounds for several learning algorithms. In particular, we prove a new margin bound for SVM, removing a log factor. The new bound is provably optimal. This resolves a long-standing open question about the PAC margin bounds achievable by SVM.
Cite
@article{arxiv.2011.04586,
title = {Stable Sample Compression Schemes: New Applications and an Optimal SVM Margin Bound},
author = {Steve Hanneke and Aryeh Kontorovich},
journal= {arXiv preprint arXiv:2011.04586},
year = {2020}
}