A Vector-Contraction Inequality for Rademacher Complexities

ALT 2016 pp. 3-17

doi:10.1007/978-3-319-46379-7_1 /alt/2016/maurer2016alt-vectorcontraction/

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.

PDF ALT Semantic Scholar

Cite

Text

Maurer. "A Vector-Contraction Inequality for Rademacher Complexities." International Conference on Algorithmic Learning Theory, 2016. doi:10.1007/978-3-319-46379-7_1

Markdown

[Maurer. "A Vector-Contraction Inequality for Rademacher Complexities." International Conference on Algorithmic Learning Theory, 2016.](https://mlanthology.org/alt/2016/maurer2016alt-vectorcontraction/) doi:10.1007/978-3-319-46379-7_1

BibTeX

@inproceedings{maurer2016alt-vectorcontraction,
  title     = {{A Vector-Contraction Inequality for Rademacher Complexities}},
  author    = {Maurer, Andreas},
  booktitle = {International Conference on Algorithmic Learning Theory},
  year      = {2016},
  pages     = {3-17},
  doi       = {10.1007/978-3-319-46379-7_1},
  url       = {https://mlanthology.org/alt/2016/maurer2016alt-vectorcontraction/}
}