Metraipolis › archive › dc_fa48b31dbf46

Beyond Mean Foils: Auditing Worst-Foil Specificity in Frozen CLIP Region Explanations — arXiv:2609.27356

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.27356
License (as archived)
CC0-1.0
Archived
2026-10-08T12:34:47.167Z
SHA-256 of the archived text
1f7bf8df56f0a7edaa885a92bec4c013f23a52467549db6f3c7885d6f92a7896
Stable URL of this page
https://metraipolis.com/documento/dc_fa48b31dbf46

How to cite

Beyond Mean Foils: Auditing Worst-Foil Specificity in Frozen CLIP Region Explanations — arXiv:2609.27356 — arXiv (en), https://arxiv.org/abs/2609.27356 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fa48b31dbf46 (Metraipolis archive dc_fa48b31dbf46, sha256 1f7bf8df56f0a7edaa885a92bec4c013f23a52467549db6f3c7885d6f92a7896)
BibTeX
@misc{dcfa48b31dbf46,
  title = {Beyond Mean Foils: Auditing Worst-Foil Specificity in Frozen CLIP Region Explanations — arXiv:2609.27356},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.27356},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fa48b31dbf46; sha256 1f7bf8df56f0a7edaa885a92bec4c013f23a52467549db6f3c7885d6f92a7896},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.27356. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

A region can overlap a target object yet contribute more to another class. We test regions selected by Cluster-based Concept Importance (CCI) in frozen CLIP. Across COCO and VOC with two checkpoints, 41.08-64.78% of regions that pass overlap and mean-contrast checks fail against the strongest competing class. Removing competitors annotated in the image leaves 39.69-63.64% failing. We then test all eight candidate regions per image. An alternative passes the test for 6.25-7.84% of failures on COCO and 27.40-31.15% on VOC. Requiring it to preserve the original target-score drop within $\epsilon = 0.02$ reduces these rates to 0.16-0.98%. Available regions and target-drop tolerance constrain repair; relaxing the tolerance increases repair opportunities.

Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-24 in cs.CV).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.27356 — 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 2026-10-08T12:12:35.668Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).