Metraipolis › archive › dc_2a9bd64ffdd5

Denoising Is Stein: Moment Access and a Resource Hierarchy for Learning Hidden Structure in Diffusion Models — arXiv:2611.04043

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.04043
License (as archived)
CC0-1.0
Archived
2026-11-04T12:39:27.574Z
SHA-256 of the archived text
317c42bbb6cee4ab141e3ca126da4d0c8d5b0e1740bf47ff1eeb5f12446455d2
Stable URL of this page
https://metraipolis.com/documento/dc_2a9bd64ffdd5

How to cite

Denoising Is Stein: Moment Access and a Resource Hierarchy for Learning Hidden Structure in Diffusion Models — arXiv:2611.04043 — arXiv (en), https://arxiv.org/abs/2611.04043 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2a9bd64ffdd5 (Metraipolis archive dc_2a9bd64ffdd5, sha256 317c42bbb6cee4ab141e3ca126da4d0c8d5b0e1740bf47ff1eeb5f12446455d2)
BibTeX
@misc{dc2a9bd64ffdd5,
  title = {Denoising Is Stein: Moment Access and a Resource Hierarchy for Learning Hidden Structure in Diffusion Models — arXiv:2611.04043},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.04043},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2a9bd64ffdd5; sha256 317c42bbb6cee4ab141e3ca126da4d0c8d5b0e1740bf47ff1eeb5f12446455d2},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.04043. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

We study score learning from denoising pairs when the data contain a hidden non-Gaussian component. If covariance reveals the component, its subspace can be recovered directly. If covariance is isotropic, the first distinguishing moment has order $k^\ast$, and the learner must extract that moment from clean points and their noisy views. Correlations in the denoising loss are Stein discrepancies against a baseline; for polynomial fields they determine exactly the corresponding moment differences. Averaging fresh noisy views conditional on a clean point gives an exact moment query on the clean distribution. In the covariance-matched regime, orientation-independent dictionaries satisfying an $L^4/L^2$ condition need order $d^{k^\ast-1}$ features, and the stated dot-product kernel ridge models need the same power in samples. Small-step online SGD on one unit has this exponent up to logarithms. A two-stage learner uses the curvature of a smoothed denoising loss to recover the hidden direction, then sharpens it and fits a scalar readout. For every even $k^\ast$, it obtains a fixed score gain over the Gaussian baseline from $O(d^{k^\ast/2}\log d)$ clean points. In the covariance-visible regime, the frozen-feature threshold occurs at $d^J/J!$, while projection onto principal components makes the fitted-model, score-risk and generated-law distributions independent of ambient dimension. In the covariance-matched experiments, the measured partial-trace overlaps are consistent with the proposed spectral transition. A width-8 network trained with clean-point reuse shows a $d^2$-like training-set threshold over $d=64$--$192$, with fixed validation and calibration overhead.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2611.04043 — 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 2026-11-04T12:29:47.266Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).