Information-Theoretic Generalization Bounds for Black-Box Learning Algorithms

Abstract

We derive information-theoretic generalization bounds for supervised learning algorithms based on the information contained in predictions rather than in the output of the training algorithm. These bounds improve over the existing information-theoretic bounds, are applicable to a wider range of algorithms, and solve two key challenges: (a) they give meaningful results for deterministic algorithms and (b) they are significantly easier to estimate. We show experimentally that the proposed bounds closely follow the generalization gap in practical scenarios for deep learning.

PDF NeurIPS OpenReview Code Semantic Scholar

Cite

Text

Harutyunyan et al. "Information-Theoretic Generalization Bounds for Black-Box Learning Algorithms." Neural Information Processing Systems, 2021.

Markdown

[Harutyunyan et al. "Information-Theoretic Generalization Bounds for Black-Box Learning Algorithms." Neural Information Processing Systems, 2021.](https://mlanthology.org/neurips/2021/harutyunyan2021neurips-informationtheoretic/)

BibTeX

@inproceedings{harutyunyan2021neurips-informationtheoretic,
  title     = {{Information-Theoretic Generalization Bounds for Black-Box Learning Algorithms}},
  author    = {Harutyunyan, Hrayr and Raginsky, Maxim and Steeg, Greg Ver and Galstyan, Aram},
  booktitle = {Neural Information Processing Systems},
  year      = {2021},
  url       = {https://mlanthology.org/neurips/2021/harutyunyan2021neurips-informationtheoretic/}
}