Which Algorithms Have Tight Generalization Bounds?

Michael Gastpar, Ido Nachum, Jonathan Shafer, Thomas Weinberger

Keywords: learning theory, overparametrization, neural networks

TL;DR: We present necessary/sufficient conditions for the existence of tight generalization bounds.

Abstract:

We study which machine learning algorithms have tight generalization bounds in the overparameterized setting. Our results build on and extend the recent work of Gastpar et al. (2024).

First, we present conditions that preclude the existence of tight generalization bounds. Specifically, we show that algorithms that have certain inductive biases that cause them to be unstable do not admit tight generalization bounds. Next, we show that algorithms that are sufficiently stable do have tight generalization bounds. We conclude with a simple characterization that relates the existence of tight generalization bounds to the conditional variance of the algorithm's loss.

Primary Area: learning theory

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 6140