Metraipolis › archive › dc_052a2ecc317a

Escaping Parameter Space: Tight Generalization Bounds via Representation Quality — arXiv:2610.23729

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.23729
License (as archived)
CC0-1.0
Archived
2026-10-19T09:10:26.503Z
SHA-256 of the archived text
aaefd59b110d77f11dbfe48d15b196ce68b3ecb32838f1cedad102d96a69be7d
Stable URL of this page
https://metraipolis.com/documento/dc_052a2ecc317a

How to cite

Escaping Parameter Space: Tight Generalization Bounds via Representation Quality — arXiv:2610.23729 — arXiv (en), https://arxiv.org/abs/2610.23729 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_052a2ecc317a (Metraipolis archive dc_052a2ecc317a, sha256 aaefd59b110d77f11dbfe48d15b196ce68b3ecb32838f1cedad102d96a69be7d)
BibTeX
@misc{dc052a2ecc317a,
  title = {Escaping Parameter Space: Tight Generalization Bounds via Representation Quality — arXiv:2610.23729},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.23729},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_052a2ecc317a; sha256 aaefd59b110d77f11dbfe48d15b196ce68b3ecb32838f1cedad102d96a69be7d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.23729. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

We derive a tight generalization bound for deterministic neural We derive a tight generalization bound for deterministic neural networks based on their last hidden layer representations. By evaluating representation quality with a parameter-free $k$-nearest-neighbor ($k$-NN) classifier, our certificate avoids traditional parameter-space complexity measures, KL divergences to a posterior, weight norms, and compression. The bound is tight: trained from scratch, it reaches an 8.0 percentage point gap to test error on CIFAR-10 (ResNet-18) and yields the first non-vacuous certificates we are aware of on CIFAR-100; with frozen DINOv2 representations, it certifies at 1.4\% on CIFAR-10 and 23.1\% on 1,000-class ImageNet. It decomposes additively into three empirical, independently measurable quantities: local class geometry ($k$-NN decoder error), decoder-network agreement, and training stability. This allows the bound to act as a diagnostic tool: we demonstrate that actively destroying local class geometry eliminates generalization while leaving dataset memorization intact, identifying class geometry as a causally relevant mechanism for generalization. Finally, we show that learned representations are significantly less sensitive to weight noise than linear readouts, suggesting why parameter-space certificates often struggle to remain tight.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.23729 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-19T09:06:29.569Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).