Metraipolis › archive › dc_30beafd9fd7e
Gaussian Limits for SGD Without Stationary Moments — arXiv:2610.05599
- Source
- arXiv
- Original
- https://arxiv.org/abs/2610.05599
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T05:32:19.384Z
- SHA-256 of the archived text
42f5c4301cc791d9a530119fb7fdb78349e6d6e286ed402c49d9a7f7d5d02f93- Stable URL of this page
https://metraipolis.com/documento/dc_30beafd9fd7e
How to cite
Gaussian Limits for SGD Without Stationary Moments — arXiv:2610.05599 — arXiv, https://arxiv.org/abs/2610.05599 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_30beafd9fd7e (Metraipolis archive dc_30beafd9fd7e, sha256 42f5c4301cc791d9a530119fb7fdb78349e6d6e286ed402c49d9a7f7d5d02f93)
BibTeX
@misc{dc30beafd9fd7e,
title = {Gaussian Limits for SGD Without Stationary Moments — arXiv:2610.05599},
howpublished = {arXiv},
url = {https://arxiv.org/abs/2610.05599},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_30beafd9fd7e; sha256 42f5c4301cc791d9a530119fb7fdb78349e6d6e286ed402c49d9a7f7d5d02f93},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.05599. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Temporal dependence can separate the Gaussian approximation of stochastic gradient descent from its stationary moments. For unmodified least-squares SGD, we construct a design with standard Gaussian marginals whose stationary error has every positive moment infinite. Independent observations with the same marginals instead give finite stationary variance. Both regimes retain a Gaussian small-step limit. Our general theory establishes pathwise contraction from a finite second design moment, then uses score cancellation and localization to obtain stationary Gaussian and Ornstein--Uhlenbeck limits. Independent Gaussian regression errors yield an exact conditional Gaussian law and total-variation convergence under the same design integrability. Stronger design conditions identify a positive first-order total-variation constant and a deterministic covariance correction with $o(a)$ error. A scalar coverage expansion translates this correction into its inference consequence. Experiments examine distributional error, coverage, and calibration with dependent scores. Together, these results establish precise probability-law approximation beyond moment-based stationary analysis. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (stat.ML) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.05599 — 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 2026-10-06T05:27:06.004Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).