Metraipolis › archive › dc_301bb0ada4e1
Beyond Local Linearity: Scale-Resolved Geometry of Learned Image Encoders — arXiv:2609.39115
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.39115
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T14:45:49.511Z
- SHA-256 of the archived text
2616e13028c65e34054f46620a9ff305e052a738159f19bdcbe762c2213fd48b- Stable URL of this page
https://metraipolis.com/documento/dc_301bb0ada4e1
How to cite
Beyond Local Linearity: Scale-Resolved Geometry of Learned Image Encoders — arXiv:2609.39115 — arXiv (en), https://arxiv.org/abs/2609.39115 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_301bb0ada4e1 (Metraipolis archive dc_301bb0ada4e1, sha256 2616e13028c65e34054f46620a9ff305e052a738159f19bdcbe762c2213fd48b)
BibTeX
@misc{dc301bb0ada4e1,
title = {Beyond Local Linearity: Scale-Resolved Geometry of Learned Image Encoders — arXiv:2609.39115},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.39115},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_301bb0ada4e1; sha256 2616e13028c65e34054f46620a9ff305e052a738159f19bdcbe762c2213fd48b},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.39115. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Understanding how learned representations respond to finite input changes is important for characterizing their sensitivity, invariances, and robustness. Yet existing geometric analyses are predominantly local and describe only infinitesimal perturbations. We introduce a scale-resolved statistic that compares an encoder's measured feature displacement with its local linear prediction as the perturbation magnitude increases. Across diverse image encoders, we discover a characteristic plateau-rise-peak-decay profile, which we call the bump. The bump is absent at initialization, emerges early during standard training, and does not form under randomized labels or random-noise inputs. Its shape also varies with the training distribution and robustness objective. These results establish departures from local geometry as a signature of how encoder representations are shaped by learning. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-10-01 in cs.CV). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2609.39115 — 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 2026-10-07T14:19:00.672Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-10-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).