Metraipolis › archive › dc_f411f4e50c10
Not All Patches Are Equally Forgettable: Spatially Localized Domain Unlearning in Vision-Language Models — arXiv:2609.17790
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.17790
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T16:16:11.791Z
- SHA-256 of the archived text
8be7ea6fe2a2e5382f0e4ccbeb298adeb7e4e2f321cc97f144826b95445991be- Stable URL of this page
https://metraipolis.com/documento/dc_f411f4e50c10
How to cite
Not All Patches Are Equally Forgettable: Spatially Localized Domain Unlearning in Vision-Language Models — arXiv:2609.17790 — arXiv (en), https://arxiv.org/abs/2609.17790 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f411f4e50c10 (Metraipolis archive dc_f411f4e50c10, sha256 8be7ea6fe2a2e5382f0e4ccbeb298adeb7e4e2f321cc97f144826b95445991be)
BibTeX
@misc{dcf411f4e50c10,
title = {Not All Patches Are Equally Forgettable: Spatially Localized Domain Unlearning in Vision-Language Models — arXiv:2609.17790},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.17790},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f411f4e50c10; sha256 8be7ea6fe2a2e5382f0e4ccbeb298adeb7e4e2f321cc97f144826b95445991be},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.17790. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Pre-trained vision-language models (VLMs) exhibit strong cross-domain recognition performance even without additional training. However, this robustness can also preserve undesirable domain-specific behavior, as domain-related and semantic information often remain entangled within the learned representation space, making selective domain unlearning challenging. Existing approaches typically address this problem through latent-space disentanglement and prompt- or feature-level interventions, without directly attributing and attenuating individual patch-token contributions. However, here we suggest that rather than uniformly suppressing the full representation, it may be more effective to exploit the spatial structure of vision transformers to localize and suppress patch regions that contribute disproportionately to forget-domain prediction. Patches that strongly influence forget-domain prediction may not be equally important for semantic recognition, suggesting that forgetting should be guided according to the domain contribution of different visual regions. Specifically, we propose a two-stage patch-selective framework that first estimates patch-level domain sensitivity and then selectively attenuates patches whose contribution to forget-domain prediction is stronger than their semantic utility. We evaluate our framework on Office-Home, Mini DomainNet, and DomainNet. Experimental results demonstrate improved forgetting-retention tradeoffs compared to prior methods while improving retained-domain recognition by up to 3.8\%. Additional evaluations under visually overlapping and unseen-domain settings further demonstrate improved robustness under distribution shift. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-17 in cs.CV). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2609.17790 — 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 2026-10-15T15:45:19.724Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-17?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).