Metraipolis › archive › dc_ca14056bc82d

Inductive Visual Logic for Few-Shot Out-Of-Distribution Adaptation in VLMs — arXiv:2609.38362

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.38362
License (as archived)
CC0-1.0
Archived
2026-10-07T14:45:13.108Z
SHA-256 of the archived text
cffd87a4a80a2765d32be1ef8158ec223a0da5faaa4a41ea4cecbf7893ef771d
Stable URL of this page
https://metraipolis.com/documento/dc_ca14056bc82d

How to cite

Inductive Visual Logic for Few-Shot Out-Of-Distribution Adaptation in VLMs — arXiv:2609.38362 — arXiv (en), https://arxiv.org/abs/2609.38362 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ca14056bc82d (Metraipolis archive dc_ca14056bc82d, sha256 cffd87a4a80a2765d32be1ef8158ec223a0da5faaa4a41ea4cecbf7893ef771d)
BibTeX
@misc{dcca14056bc82d,
  title = {Inductive Visual Logic for Few-Shot Out-Of-Distribution Adaptation in VLMs — arXiv:2609.38362},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.38362},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ca14056bc82d; sha256 cffd87a4a80a2765d32be1ef8158ec223a0da5faaa4a41ea4cecbf7893ef771d},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.38362. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Generative vision-language models (VLMs) such as Qwen-VL and LLaVA achieve strong zero-shot performance on tasks overlapping with their pretraining distribution, yet fail on specialized domains where the required discriminative features were never learned, a regime we term distant out-of-distribution (OOD). Standard adaptation methods cannot overcome this representational absence because they operate within the encoder's existing feature space. However, VLMs retain a robust descriptive capacity even when discrimination collapses: a model that cannot classify a medical scan can still articulate its visual patterns. Exploiting this asymmetry, we introduce Inductive Visual Logic (IVL), a training-free framework that constructs classification knowledge from the model's surviving descriptive ability. IVL extracts visual traits from few-shot support images through dual-mode prompting, combining semantic descriptions with primitive visual observations, and organizes them into per-class trait dictionaries. At inference, hierarchical filtering identifies spatially grounded trait evidence for classification. Across multiple distant-OOD benchmarks, IVL achieves the highest aggregate accuracy under two VLM backbones while producing interpretable, trait-traceable predictions.

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/2609.38362 — 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 2026-10-07T14:19:00.543Z 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).