A vector-contraction inequality for Rademacher complexities
Machine Learning
2016-05-04 v1 Machine Learning
Abstract
The contraction inequality for Rademacher averages is extended to Lipschitz functions with vector-valued domains, and it is also shown that in the bounding expression the Rademacher variables can be replaced by arbitrary iid symmetric and sub-gaussian variables. Example applications are given for multi-category learning, K-means clustering and learning-to-learn.
Keywords
Cite
@article{arxiv.1605.00251,
title = {A vector-contraction inequality for Rademacher complexities},
author = {Andreas Maurer},
journal= {arXiv preprint arXiv:1605.00251},
year = {2016}
}