ML Anthology

Constant Rate Approximate Maximum Margin Algorithms

Petroula Tsampouka, John Shawe-Taylor
ECML-PKDD 2006 pp. 437-448
doi:10.1007/11871842_42 /ecmlpkdd/2006/tsampouka2006ecml-constant/

Abstract

We present a new class of Perceptron-like algorithms with margin in which the “effective” learning rate η _eff, defined as the ratio of the learning rate to the length of the weight vector, remains constant. We prove that for η _eff sufficiently small the new algorithms converge in a finite number of steps and show that there exists a limit of the parameters involved in which convergence leads to classification with maximum margin. A soft margin extension for Perceptron-like large margin classifiers is also discussed.

PDF ECML-PKDD Semantic Scholar

Cite

Text

Tsampouka and Shawe-Taylor. "Constant Rate Approximate Maximum Margin Algorithms." European Conference on Machine Learning, 2006. doi:10.1007/11871842_42

Markdown

[Tsampouka and Shawe-Taylor. "Constant Rate Approximate Maximum Margin Algorithms." European Conference on Machine Learning, 2006.](https://mlanthology.org/ecmlpkdd/2006/tsampouka2006ecml-constant/) doi:10.1007/11871842_42

BibTeX

@inproceedings{tsampouka2006ecml-constant,
  title     = {{Constant Rate Approximate Maximum Margin Algorithms}},
  author    = {Tsampouka, Petroula and Shawe-Taylor, John},
  booktitle = {European Conference on Machine Learning},
  year      = {2006},
  pages     = {437-448},
  doi       = {10.1007/11871842_42},
  url       = {https://mlanthology.org/ecmlpkdd/2006/tsampouka2006ecml-constant/}
}