Seeing to Ground: Visual Attention for Hallucination-Resilient MDLLMs

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Narnaware, Vishal, Gupta, Animesh, Zhai, Kevin, Wang, Zhenyi, Shah, Mubarak
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866910076086779904
author Narnaware, Vishal
Gupta, Animesh
Zhai, Kevin
Wang, Zhenyi
Shah, Mubarak
author_facet Narnaware, Vishal
Gupta, Animesh
Zhai, Kevin
Wang, Zhenyi
Shah, Mubarak
contents Multimodal Diffusion Large Language Models (MDLLMs) achieve high-concurrency generation through parallel masked decoding, yet the architectures remain prone to multimodal hallucinations. This structural vulnerability stems from an algorithmic flaw: the decoder ranks candidate tokens based on textual likelihood without verifying localized visual support. We establish that this language-only ranking induces an objective mismatch, where language probability mass acts as a misspecified proxy for the intended multimodal task. Consequently, we reinterpret hallucination as a localized optimization error, a phenomenon where the decoder exploits language shortcuts to maximize a proxy score at the expense of visual grounding. To address this objective mismatch, we introduce VISAGE, a training-free decoding framework that calibrates the objective at inference time. VISAGE estimates the proxy discrepancy by quantifying the spatial entropy of cross-attention distributions. By enforcing a localization consensus across attention heads, the method penalizes spatially uniform distributions and re-ranks token commitments to favor visually grounded outcomes. We provide an analytical stability guarantee establishing that VISAGE maintains a bounded objective loss under estimation error. Evaluations across hallucination-sensitive and general-purpose benchmarks demonstrate the robustness of the framework, yielding relative gains of 8.59% on MMMU-val and 7.75% on HallusionBench.
format Preprint
id arxiv_https___arxiv_org_abs_2603_25711
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Seeing to Ground: Visual Attention for Hallucination-Resilient MDLLMs
Narnaware, Vishal
Gupta, Animesh
Zhai, Kevin
Wang, Zhenyi
Shah, Mubarak
Computer Vision and Pattern Recognition
Multimodal Diffusion Large Language Models (MDLLMs) achieve high-concurrency generation through parallel masked decoding, yet the architectures remain prone to multimodal hallucinations. This structural vulnerability stems from an algorithmic flaw: the decoder ranks candidate tokens based on textual likelihood without verifying localized visual support. We establish that this language-only ranking induces an objective mismatch, where language probability mass acts as a misspecified proxy for the intended multimodal task. Consequently, we reinterpret hallucination as a localized optimization error, a phenomenon where the decoder exploits language shortcuts to maximize a proxy score at the expense of visual grounding. To address this objective mismatch, we introduce VISAGE, a training-free decoding framework that calibrates the objective at inference time. VISAGE estimates the proxy discrepancy by quantifying the spatial entropy of cross-attention distributions. By enforcing a localization consensus across attention heads, the method penalizes spatially uniform distributions and re-ranks token commitments to favor visually grounded outcomes. We provide an analytical stability guarantee establishing that VISAGE maintains a bounded objective loss under estimation error. Evaluations across hallucination-sensitive and general-purpose benchmarks demonstrate the robustness of the framework, yielding relative gains of 8.59% on MMMU-val and 7.75% on HallusionBench.
title Seeing to Ground: Visual Attention for Hallucination-Resilient MDLLMs
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.25711