Metraipolis › archive › dc_bd91d99fca09

Koopman Observers for Diffusion Acceleration: Correcting Feature Forecasts with Shallow Measurements — arXiv:2610.10366

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.10366
License (as archived): CC0-1.0
Archived: 2026-10-08T12:36:23.083Z
SHA-256 of the archived text: 5896bac153eb241917016020480bcb49383d43a5f91bd54ac818d9cb9cbbb9d2
Stable URL of this page: https://metraipolis.com/documento/dc_bd91d99fca09

How to cite

Koopman Observers for Diffusion Acceleration: Correcting Feature Forecasts with Shallow Measurements — arXiv:2610.10366 — arXiv (en), https://arxiv.org/abs/2610.10366 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bd91d99fca09 (Metraipolis archive dc_bd91d99fca09, sha256 5896bac153eb241917016020480bcb49383d43a5f91bd54ac818d9cb9cbbb9d2)

BibTeX

@misc{dcbd91d99fca09,
  title = {Koopman Observers for Diffusion Acceleration: Correcting Feature Forecasts with Shallow Measurements — arXiv:2610.10366},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.10366},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bd91d99fca09; sha256 5896bac153eb241917016020480bcb49383d43a5f91bd54ac818d9cb9cbbb9d2},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.10366. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Feature caching accelerates diffusion sampling by replacing expensive network evaluations with predictions from previously computed activations. However, forecasts based only on past features cannot directly incorporate changes in the current denoising state. We investigate whether inexpensive, freshly computed features can serve as observations for correcting these predictions. We introduce an observation-corrected Koopman framework for accelerating frozen diffusion models. Using calibration trajectories, we identify finite-dimensional, time-dependent Koopman approximations that jointly describe the increments of shallow and deep network features. During accelerated sampling, these operators predict the evolution of expensive deep features, while innovations in the observed shallow features correct the predicted state. Periodic full evaluations refresh the observer, and all generative-model parameters remain unchanged. This formulation enables controlled comparisons of temporal prediction and observation correction. Across three 10,000-image runs per dataset, our method reduces paired Inception-feature MSE by $19.9\%$ on CIFAR-10 and $11.9\%$ on a ten-class ImageNet subset relative to channelwise affine prediction under the same four-partial-step schedule. Matched ablations attribute additional reductions of $4.54\%$ and $4.67\%$ to observation correction. The observer achieves $1.89\times$ and $1.85\times$ measured speedups over DDIM-50, supporting improved reference-sampler fidelity without retraining the denoiser.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG) ; Dynamical Systems (math.DS); Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2610.10366 — 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 2026-10-08T12:14:15.338Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).