Characterization of Overfitting in Robust Multiclass Classification

Abstract

This paper considers the following question: Given the number of classes m, the number of robust accuracy queries k, and the number of test examples in the dataset n, how much can adaptive algorithms robustly overfit the test dataset? We solve this problem by equivalently giving near-matching upper and lower bounds of the robust overfitting bias in multiclass classification problems.

PDF NeurIPS OpenReview Semantic Scholar

Cite

Text

Xu and Liu. "Characterization of Overfitting in Robust Multiclass Classification." Neural Information Processing Systems, 2023.

Markdown

[Xu and Liu. "Characterization of Overfitting in Robust Multiclass Classification." Neural Information Processing Systems, 2023.](https://mlanthology.org/neurips/2023/xu2023neurips-characterization/)

BibTeX

@inproceedings{xu2023neurips-characterization,
  title     = {{Characterization of Overfitting in Robust Multiclass Classification}},
  author    = {Xu, Jingyuan and Liu, Weiwei},
  booktitle = {Neural Information Processing Systems},
  year      = {2023},
  url       = {https://mlanthology.org/neurips/2023/xu2023neurips-characterization/}
}