Saved in:
Bibliographic Details
Main Authors: Chen, Shirui, Recanatesi, Stefano, Shea-Brown, Eric
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2310.01770
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911461581783040
author Chen, Shirui
Recanatesi, Stefano
Shea-Brown, Eric
author_facet Chen, Shirui
Recanatesi, Stefano
Shea-Brown, Eric
contents Despite extensive study, the significance of sharpness -- the trace of the loss Hessian at local minima -- remains unclear. We investigate an alternative perspective: how sharpness relates to the geometric structure of neural representations, specifically representation compression, defined as how strongly neural activations concentrate under local input perturbations. We introduce three measures -- Local Volumetric Ratio (LVR), Maximum Local Sensitivity (MLS), and Local Dimensionality -- and derive upper bounds showing these are mathematically constrained by sharpness: flatter minima necessarily limit compression. We extend these bounds to reparametrization-invariant sharpness and introduce network-wide variants (NMLS, NVR) that provide tighter, more stable bounds than prior single-layer analyses. Empirically, we validate consistent positive correlations across feedforward, convolutional, and transformer architectures. Our results suggest that sharpness fundamentally quantifies representation compression, offering a principled resolution to contradictory findings on the sharpness-generalization relationship.
format Preprint
id arxiv_https___arxiv_org_abs_2310_01770
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle A simple connection from loss flatness to compressed neural representations
Chen, Shirui
Recanatesi, Stefano
Shea-Brown, Eric
Machine Learning
Artificial Intelligence
Despite extensive study, the significance of sharpness -- the trace of the loss Hessian at local minima -- remains unclear. We investigate an alternative perspective: how sharpness relates to the geometric structure of neural representations, specifically representation compression, defined as how strongly neural activations concentrate under local input perturbations. We introduce three measures -- Local Volumetric Ratio (LVR), Maximum Local Sensitivity (MLS), and Local Dimensionality -- and derive upper bounds showing these are mathematically constrained by sharpness: flatter minima necessarily limit compression. We extend these bounds to reparametrization-invariant sharpness and introduce network-wide variants (NMLS, NVR) that provide tighter, more stable bounds than prior single-layer analyses. Empirically, we validate consistent positive correlations across feedforward, convolutional, and transformer architectures. Our results suggest that sharpness fundamentally quantifies representation compression, offering a principled resolution to contradictory findings on the sharpness-generalization relationship.
title A simple connection from loss flatness to compressed neural representations
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2310.01770