English

Domain based classification

Machine Learning 2018-08-14 v2 Machine Learning

Abstract

The majority of traditional classification ru les minimizing the expected probability of error (0-1 loss) are inappropriate if the class probability distributions are ill-defined or impossible to estimate. We argue that in such cases class domains should be used instead of class distributions or densities to construct a reliable decision function. Proposals are presented for some evaluation criteria and classifier learning schemes, illustrated by an example.

Keywords

Cite

@article{arxiv.1601.04530,
  title  = {Domain based classification},
  author = {Robert P. W. Duin and Elzbieta Pekalska},
  journal= {arXiv preprint arXiv:1601.04530},
  year   = {2018}
}

Comments

8 pages, unpublished paper written in 2005, discussing a significant, still almost not studied problem. missing reference links corrected

R2 v1 2026-06-22T12:31:43.722Z