Metraipolis › archive › dc_2e83e75a6006
SurvAD: How Anomalies Survive Diffusion Reconstruction — arXiv:2611.09997
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.09997
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-10T18:42:06.604Z
- SHA-256 of the archived text
4ce8f6aa5001d699cf7aa3bec0a21bc859901fcfb249d2a53b1cde79517f1fd3- Stable URL of this page
https://metraipolis.com/documento/dc_2e83e75a6006
How to cite
SurvAD: How Anomalies Survive Diffusion Reconstruction — arXiv:2611.09997 — arXiv (en), https://arxiv.org/abs/2611.09997 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2e83e75a6006 (Metraipolis archive dc_2e83e75a6006, sha256 4ce8f6aa5001d699cf7aa3bec0a21bc859901fcfb249d2a53b1cde79517f1fd3)
BibTeX
@misc{dc2e83e75a6006,
title = {SurvAD: How Anomalies Survive Diffusion Reconstruction — arXiv:2611.09997},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.09997},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2e83e75a6006; sha256 4ce8f6aa5001d699cf7aa3bec0a21bc859901fcfb249d2a53b1cde79517f1fd3},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.09997. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Diffusion-based anomaly detectors reconstruct a test image towards normality, yet how much of an anomaly survives, its leak, remains uncharacterized. We show that a first-order deterministic sampler obeys an exact recursion along the anomaly, whose continuous limit is a survival law. The anomaly decays exponentially as the noise falls, at a rate set by the retention $r$, the share the model keeps on the states it visits. Because $r$ depends on the noise level and not on the step count, one measured path predicts the leak of other step counts, samplers and schedules from a nearby start to within a few hundredths, on four diffusion models and real defects. The law also shows that the leak's growth over steps is exposure bias and that normal details obey the same law, so changing the path or the training trades leak for detail error. We introduce SurvAD, which fine-tunes the reconstructor on inference-like starts and fits a comparator that adds evidence to the residual map without suppressing its ranking elsewhere. For under $2\%$ more inference time, SurvAD raises the multi-class pixel average precision (P-AP) of DeCo-Diff from $76.8$ to $80.7$ on MVTec AD and from $57.5$ to $60.0$ on VisA. It matches CostFilter-AD on the same reconstructor with $9\times$ fewer trainable parameters and, under Dinomaly's protocol, exceeds its reported P-AP on Dinomaly on both benchmarks. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2611.09997 — 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 2026-11-10T18:37:37.564Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).