Metraipolis › archive › dc_f488d01ababe

AIM: Anchor Identity Features, Then Match for Multimodal Large Language Model Unlearning — arXiv:2608.28312

Source: arXiv (en)
Original: https://arxiv.org/abs/2608.28312
License (as archived): CC0-1.0
Archived: 2026-10-17T10:56:45.273Z
SHA-256 of the archived text: 86428532c0485988378634ded99298c729631c5b892093a6f98ec22284897f3a
Stable URL of this page: https://metraipolis.com/documento/dc_f488d01ababe

How to cite

AIM: Anchor Identity Features, Then Match for Multimodal Large Language Model Unlearning — arXiv:2608.28312 — arXiv (en), https://arxiv.org/abs/2608.28312 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f488d01ababe (Metraipolis archive dc_f488d01ababe, sha256 86428532c0485988378634ded99298c729631c5b892093a6f98ec22284897f3a)

BibTeX

@misc{dcf488d01ababe,
  title = {AIM: Anchor Identity Features, Then Match for Multimodal Large Language Model Unlearning — arXiv:2608.28312},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.28312},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f488d01ababe; sha256 86428532c0485988378634ded99298c729631c5b892093a6f98ec22284897f3a},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.28312. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Multimodal large language models (MLLMs) can memorize identity-specific facts about people in their fine-tuning data, creating privacy risks when a person requests deletion. Existing MLLM unlearning methods often assume access to retain images or ground-truth answers during deletion, which is unrealistic in many practical scenarios. We study identity unlearning when retain images are unavailable at deletion time. Our analysis shows that identity and visual-perception questions occupy distinct regions in fine-tuned hidden states and are organized differently: identity questions cluster by person, whereas perception questions cluster by question type. This suggests that identity knowledge can be suppressed without erasing general visual perception. Building on this observation, we propose AIM, a two-stage method that anchors an identity-forgetting target with a universal visual prompt and then matches the vision encoder to that target under a Fisher-based constraint. Extensive experiments show that AIM achieves competitive identity forgetting while preserving non-deleted identities, prior knowledge, and visual perception on the same images.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-31 in cs.CL).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.28312 — 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 2026-10-17T10:47:33.166Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-31?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).