Metraipolis › archive › dc_fef8cd5c814e

Feature Selective Model Collapse in Diffusion Models: Total Replacement versus Fixed-Budget Training — arXiv:2610.01318

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.01318
License (as archived)
CC0-1.0
Archived
2026-10-06T12:13:32.853Z
SHA-256 of the archived text
3cc49bb7e0bf72a1b71ea56c601255754d4568094107e4d14a8cc0681ff0d20a
Stable URL of this page
https://metraipolis.com/documento/dc_fef8cd5c814e

How to cite

Feature Selective Model Collapse in Diffusion Models: Total Replacement versus Fixed-Budget Training — arXiv:2610.01318 — arXiv (en), https://arxiv.org/abs/2610.01318 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fef8cd5c814e (Metraipolis archive dc_fef8cd5c814e, sha256 3cc49bb7e0bf72a1b71ea56c601255754d4568094107e4d14a8cc0681ff0d20a)
BibTeX
@misc{dcfef8cd5c814e,
  title = {Feature Selective Model Collapse in Diffusion Models: Total Replacement versus Fixed-Budget Training — arXiv:2610.01318},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.01318},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fef8cd5c814e; sha256 3cc49bb7e0bf72a1b71ea56c601255754d4568094107e4d14a8cc0681ff0d20a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.01318. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Model collapse arises when generative models are trained on synthetic data produced by earlier models. The phenomenon has attracted considerable attention because of its societal and technical implications. However, previous studies have reached seemingly contradictory conclusions: replacing real data with synthetic data causes collapse (Shumailov et al.), yet accumulating real data alongside synthetic data can prevent it. For diffusion models, we study an intermediate regime typical of finite-budget pipelines: all past datasets and the real data are kept, but each new model is trained on a fixed-size sample from this growing pool, so the real fraction vanishes without any data being removed. Experiments on a 2D spiral dataset as well as the image benchmarks (MNIST, Fashion-MNIST, and CIFAR-10) show that replacement protocol degrades dataset rapidly as in the literature, whereas the fixed budget degrades only partially, sparing some features. A linear-response model of the multi-generational parameter dynamics, analyzed by stochastic recursion, confirms that the two protocols differ: some features will be fragile and lost within a few generations for both protocols, while some will be robust and preserved over practically unbounded horizons under the fixed budget protocol.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-10-02 in cs.AI).
Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI); Statistics Theory (math.ST).
Pagina del lavoro: https://arxiv.org/abs/2610.01318 — 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 2026-10-06T10:21:28.176Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-10-02?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).