On Uniform Convergence and Low-Norm Interpolation Learning

Authors: Lijia Zhou, Danica J. Sutherland, Nati Srebro

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We consider an underdetermined noisy linear regression model where the minimum-norm interpolating predictor is known to be consistent, and ask: can uniform convergence in a norm ball, or at least (following Nagarajan and Kolter) the subset of a norm ball that the algorithm selects on a typical input set, explain this success? We show that uniformly bounding the difference between empirical and population errors cannot show any learning in the norm ball, and cannot show consistency for any set, even one depending on the exact algorithm and distribution. But we argue we can explain the consistency of the minimal-norm interpolator with a slightly weaker, yet standard, notion: uniform convergence of zero-error predictors in a norm ball. We use this to bound the generalization error of low(but not minimal-) norm interpolating predictors.
Researcher Affiliation Academia Lijia Zhou University of Chicago zlj@uchicago.edu Danica J. Sutherland TTI-Chicago danica@ttic.edu Nathan Srebro TTI-Chicago nati@ttic.edu
Pseudocode No The paper is theoretical and focuses on mathematical derivations and proofs, not pseudocode or algorithm blocks.
Open Source Code No The paper does not mention providing open-source code for its methodology.
Open Datasets No The paper is theoretical and does not mention using any datasets for training.
Dataset Splits No The paper is theoretical and does not specify any dataset splits for validation or training.
Hardware Specification No The paper is theoretical and does not mention any hardware specifications used for experiments.
Software Dependencies No The paper is theoretical and does not mention any specific software dependencies with version numbers.
Experiment Setup No The paper is theoretical and does not describe any experimental setup or hyperparameters.