Modelling Class Noise with Symmetric and Asymmetric Distributions

Abstract

In classification problem, we assume that the samples around the class boundary are more likely to be incorrectly annotated than others, and propose boundary-conditional class noise (BCN). Based on the BCN assumption, we use unnormalized Gaussian and Laplace distributions to directly model how class noise is generated, in symmetric and asymmetric cases. In addition, we demonstrate that Logistic regression and Probit regression can also be reinterpreted from this class noise perspective, and compare them with the proposed models. The empirical study shows that, the proposed asymmetric models overall outperform the benchmark linear models, and the asymmetric Laplace-noise model achieves the best performance among all.

Cite

Text

Du and Cai. "Modelling Class Noise with Symmetric and Asymmetric Distributions." AAAI Conference on Artificial Intelligence, 2015. doi:10.1609/AAAI.V29I1.9612

Markdown

[Du and Cai. "Modelling Class Noise with Symmetric and Asymmetric Distributions." AAAI Conference on Artificial Intelligence, 2015.](https://mlanthology.org/aaai/2015/du2015aaai-modelling/) doi:10.1609/AAAI.V29I1.9612

BibTeX

@inproceedings{du2015aaai-modelling,
  title     = {{Modelling Class Noise with Symmetric and Asymmetric Distributions}},
  author    = {Du, Jun and Cai, Zhihua},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {2015},
  pages     = {2589-2595},
  doi       = {10.1609/AAAI.V29I1.9612},
  url       = {https://mlanthology.org/aaai/2015/du2015aaai-modelling/}
}