No AI summary available for this article.
Why It Matters
Removing a specific individual's information from multimodal large language models (MLLMs) is often needed after deployment, but existing methods rely on a retain set, which is hardest to obtain at that point, and rebuilding it recreates the privacy exposure that unlearning aims to remove.
Provenance
Discovered via ArXiv and published by ArXiv.
Key Claims
Original description
Removing a specific individual's information from multimodal large language models (MLLMs) is often needed after deployment, but existing methods rely on a retain set, which is hardest to obtain at that point, and rebuilding it recreates the privacy exposure that unlearning aims to remove. Forgetting from the forget set alone instead damages the shared visual-language computation, harming perception. We cast retain-free unlearning as a localization problem: causal tracing, weight transplant, and Fisher overlap all point to early-to-mid decoder MLPs as the layers where identity information is s...
Discovered via ArXiv
Research papers and preprints from arXiv.
Publisher: arxiv.org
ID: http://arxiv.org/abs/2608.30649v1 · Indexed 1 day ago