arXiv AI By Jiaao Yu, Yujian Ma, Xianming Hu, Pengran Wang, Ang Li

Training-Free VLM Personalization via Calibrated Residual Decoding

Read the original on arXiv AI →

The paper introduces a training‑free method for personalizing vision‑language models by using calibrated residual decoding. It constructs three evidence conditions—positive, counterfactual, and empty profiles—to isolate the true contribution of personalization. Normalized‑entropy calibration adjusts the influence of this residual signal based on its uncertainty, improving identity‑sensitive visual personalization without fine‑tuning.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

Hugging Face Trending Papers
Jun 14

Mitigating Visual Hallucinations in Multimodal Systems through Retrieval-Augmented Reliability-Aware Inference

Multimodal large language models (MLLMs) have demonstrated strong capabilities in vision-language understanding and natural-language response generation. However, these systems can still produce overconfident predictions and hallucination-like outputs, particularly when the visual evidence is weak, ambiguous, or semantically inconsistent.

arXiv AI
Sep 7

When Seeing Overrides Knowing: Visual Dominance and Deferral-Based Method for Personalized Safety in VLMs

The paper introduces MPS-Bench, a benchmark of 5,181 scenarios from 584 real-world images across 12 high-risk domains, each paired with a hidden user profile, to evaluate personalized safety in vision‑language models (VLMs). Eight leading VLMs were tested and found to almost always respond directly (86‑99%) without seeking missing context, scoring no higher than 2.6/5 on personalized safety. The authors identify a phenomenon called visual dominance, where visual information enters text representations early and suppresses textual risk signals, and propose PRISM, a lightweight input monitor that predicts when a query should be deferred, achieving 0.978 AUC and outperforming all tested models on the safety‑utility Pareto frontier.

By Edward Sun, Yuchen Wu, Zixian Ma, Eric Hanchen Jiang, Yijia Xiao, Xiaoyuan Yi, Ranjay Krishna, Wei Wang, Jindong Wang, Aylin Caliskan