Skip to content

Latest commit

 

History

History
20 lines (15 loc) · 1.22 KB

README.md

File metadata and controls

20 lines (15 loc) · 1.22 KB

$DOC^3$ - Deep One Class Classifiction using Contradictions

This is the accompanying code for the paper published in the Machine Learning journal in 2023. In this work we introduce the notion of learning from contradictions (a.k.a. Universum Learning) for deep one class classification problems. We formalize this notion for the widely adopted one class large-margin loss, and propose the Deep One Class Classification using Contradictions ($DOC^3$) algorithm. We show that learning from contradictions incurs lower generalization error by comparing the Empirical Rademacher Complexity (ERC) of $DOC^3$ against its traditional inductive learning counterpart. Our empirical results demonstrate the efficacy of $DOC^3$ compared to popular baseline algorithms on several real-life data sets.

If you find this work useful please cite,

here goes the reference to our paper

Dependencies

  • python 3.9.16
  • pytorch 2.0.0
  • numpy 1.23.5
  • matplotlib 3.7.1
  • scikit-learn 1.2.0
  • torchvision 0.15.0

License

$DOC^3$ is open-sourced under the Apache-2.0 license. See the LICENSE file for details.

Contact

Bernardo Gonzalez <[email protected]>