Worst-Case Quadratic Loss Bounds for a Generalization of the Widrow-Hoff Rule

Nicolò Cesa-Bianchi, Philip M. Long, Manfred K. Warmuth

COLT 1993 pp. 429-438

doi:10.1145/168304.168390 /colt/1993/cesabianchi1993colt-worst/

Abstract

We prove worst-case bounds on the sum of squared errors incurred by a generalization of the classical Widrow-Hoff algorithm to inner product spaces. We describe applications of this result to obtain worst-case agnostic learning results for classes of smooth functions and of linear functions.

COLT Semantic Scholar

Cite

Text

Cesa-Bianchi et al. "Worst-Case Quadratic Loss Bounds for a Generalization of the Widrow-Hoff Rule." Annual Conference on Computational Learning Theory, 1993. doi:10.1145/168304.168390

Markdown

[Cesa-Bianchi et al. "Worst-Case Quadratic Loss Bounds for a Generalization of the Widrow-Hoff Rule." Annual Conference on Computational Learning Theory, 1993.](https://mlanthology.org/colt/1993/cesabianchi1993colt-worst/) doi:10.1145/168304.168390

BibTeX

@inproceedings{cesabianchi1993colt-worst,
  title     = {{Worst-Case Quadratic Loss Bounds for a Generalization of the Widrow-Hoff Rule}},
  author    = {Cesa-Bianchi, Nicolò and Long, Philip M. and Warmuth, Manfred K.},
  booktitle = {Annual Conference on Computational Learning Theory},
  year      = {1993},
  pages     = {429-438},
  doi       = {10.1145/168304.168390},
  url       = {https://mlanthology.org/colt/1993/cesabianchi1993colt-worst/}
}