Metraipolis › archive › dc_d80d09b65a2b

Clipped or Unclipped? Finite-Sample Trade-offs for Averaged SGD under Heavy-Tailed Noise — arXiv:2609.32962

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.32962
License (as archived)
CC0-1.0
Archived
2026-10-07T22:37:02.339Z
SHA-256 of the archived text
b2d6f3b691e458a6f383f1fb296f715e54874ad8d2d559d142a29fcd1d67d0e8
Stable URL of this page
https://metraipolis.com/documento/dc_d80d09b65a2b

How to cite

Clipped or Unclipped? Finite-Sample Trade-offs for Averaged SGD under Heavy-Tailed Noise — arXiv:2609.32962 — arXiv (en), https://arxiv.org/abs/2609.32962 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d80d09b65a2b (Metraipolis archive dc_d80d09b65a2b, sha256 b2d6f3b691e458a6f383f1fb296f715e54874ad8d2d559d142a29fcd1d67d0e8)
BibTeX
@misc{dcd80d09b65a2b,
  title = {Clipped or Unclipped? Finite-Sample Trade-offs for Averaged SGD under Heavy-Tailed Noise — arXiv:2609.32962},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.32962},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d80d09b65a2b; sha256 b2d6f3b691e458a6f383f1fb296f715e54874ad8d2d559d142a29fcd1d67d0e8},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32962. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Gradient clipping is widely used to stabilize training, but it need not improve the statistical accuracy of averaged SGD, even under heavy-tailed noise. We derive a finite-sample comparison of clipped and unclipped Polyak-Ruppert averaged SGD under finite conditional $p$-th moments, $p\ge2$. Our main result gives explicit accuracy and confidence conditions under which, for $p>2$, the Gaussian term dominates the unclipped deviation bound, so clipping need not improve its leading order. By balancing clipping bias and concentration, we obtain a bound in which the heavy-tail correction depends logarithmically rather than polynomially on the inverse failure probability. At $p=2$, this improves the confidence dependence of the leading bound. We establish sharpness of the unclipped heavy-tail term through an exact one-dimensional quadratic recursion and extend the comparison to projected convex SGD. We also prove concrete costs of clipping: every fixed finite threshold increases asymptotic variance on a scalar Gaussian quadratic, while whole-gradient clipping can shift the limiting point under asymmetric noise.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.LG).
Categorie arXiv: Machine Learning (cs.LG) ; Optimization and Control (math.OC).
Pagina del lavoro: https://arxiv.org/abs/2609.32962 — 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 2026-10-07T22:14:27.827Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).