Copertina

When Looking Is Not Enough: Visual Attention Structure Reveals Hallucination in MLLMs

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Cao, Fanpu, Zou, Xin, Hu, Xuming, Xiong, Hui
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Computer Vision and Pattern Recognition Artificial Intelligence
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Vision-Language Introspection: Mitigating Overconfident Hallucinations in MLLMs via Interpretable Bi-Causal Steering
di: Liu, Shuliang, et al.
Pubblicazione: (2026)

MLLMs Know When Before Speaking: Revealing and Recovering Temporal Grounding via Attention Cues
di: Du, Dazhao, et al.
Pubblicazione: (2026)

Decoding by Perturbation: Mitigating MLLM Hallucinations via Dynamic Textual Perturbation
di: Jia, Sihang, et al.
Pubblicazione: (2026)

Mitigating Visual Hallucinations via Semantic Curriculum Preference Optimization in MLLMs
di: Li, Yuanshuai, et al.
Pubblicazione: (2025)

Sharp Eyes and Memory for VideoLLMs: Information-Aware Visual Token Pruning for Efficient and Reliable VideoLLM Reasoning
di: Qin, Jialong, et al.
Pubblicazione: (2025)

When One Moment Isn't Enough: Multi-Moment Retrieval with Cross-Moment Interactions
di: Cao, Zhuo, et al.
Pubblicazione: (2025)

StemBind: When MLLMs Get Lost Between Rules and Instances in Abstract Visual Reasoning
di: He, Xixiang, et al.
Pubblicazione: (2026)

Explore the Hallucination on Low-level Perception for MLLMs
di: Sun, Yinan, et al.
Pubblicazione: (2024)

Do MLLMs Really See It: Reinforcing Visual Attention in Multimodal LLMs
di: Ou, Siqu, et al.
Pubblicazione: (2026)

Multimodal Language Models See Better When They Look Shallower
di: Chen, Haoran, et al.
Pubblicazione: (2025)

FINER: MLLMs Hallucinate under Fine-grained Negative Queries
di: Xiao, Rui, et al.
Pubblicazione: (2026)

Thinking Before Looking: Improving Multimodal LLM Reasoning via Mitigating Visual Hallucination
di: Zheng, Haojie, et al.
Pubblicazione: (2024)

CrystaL: Spontaneous Emergence of Visual Latents in MLLMs
di: Zhang, Yang, et al.
Pubblicazione: (2026)

MLLMs Know Where to Look: Training-free Perception of Small Visual Details with Multimodal LLMs
di: Zhang, Jiarui, et al.
Pubblicazione: (2025)

Tell Model Where to Look: Mitigating Hallucinations in MLLMs by Vision-Guided Attention
di: Zhao, Jianfei, et al.
Pubblicazione: (2025)

LookWise: Knowing When and Where to Look for Fine-Grained Visual Reasoning in Multimodal Large Language Models
di: Shen, Yuxiang, et al.
Pubblicazione: (2026)

Moment-Video: Diagnosing Temporal Fidelity of Video MLLMs on Momentary Visual Events
di: Liu, Xiaolin, et al.
Pubblicazione: (2026)

Looking Back and Forth: Cross-Image Attention Calibration and Attentive Preference Learning for Multi-Image Hallucination Mitigation
di: Yang, Xiaochen, et al.
Pubblicazione: (2026)

Taming Hallucinations: Boosting MLLMs' Video Understanding via Counterfactual Video Generation
di: Huang, Zhe, et al.
Pubblicazione: (2025)

StrLoRA: Towards Streaming Continual Visual Instruction Tuning for MLLMs
di: Che, Chang, et al.
Pubblicazione: (2026)

Look Before You Decide: Prompting Active Deduction of MLLMs for Assumptive Reasoning
di: Li, Yian, et al.
Pubblicazione: (2024)

Mitigating Object Hallucinations in MLLMs via Multi-Frequency Perturbations
di: Li, Shuo, et al.
Pubblicazione: (2025)

Attention at Rest Stays at Rest: Breaking Visual Inertia for Cognitive Hallucination Mitigation
di: Gong, Boyang, et al.
Pubblicazione: (2026)

Correcting Visual Blur Induced by Attention Distraction to Reduce Hallucinations: Algorithm and Theory
di: Li, Quanjiang, et al.
Pubblicazione: (2026)

Global Context or Local Detail? Adaptive Visual Grounding for Hallucination Mitigation
di: Jiang, Yubo, et al.
Pubblicazione: (2026)

VSI: Visual Subtitle Integration for Keyframe Selection to enhance Long Video Understanding
di: He, Jianxiang, et al.
Pubblicazione: (2025)

Finding the Correct Visual Evidence Without Forgetting: Mitigating Hallucination in LVLMs via Inter-Layer Visual Attention Discrepancy
di: Xie, Yutong, et al.
Pubblicazione: (2026)

When Few Steps Are Enough: Training-Free Acceleration of Identity-Preserved Generation
di: Zheng, Dongqi
Pubblicazione: (2026)

S$^2$-MLLM: Boosting Spatial Reasoning Capability of MLLMs for 3D Visual Grounding with Structural Guidance
di: Xu, Beining, et al.
Pubblicazione: (2025)

Batch Transformer: Look for Attention in Batch
di: Her, Myung Beom, et al.
Pubblicazione: (2024)

Beyond Attention or Similarity: Maximizing Conditional Diversity for Token Pruning in MLLMs
di: Zhang, Qizhe, et al.
Pubblicazione: (2025)

HiDe: Rethinking The Zoom-IN method in High Resolution MLLMs via Hierarchical Decoupling
di: Liu, Xianjie, et al.
Pubblicazione: (2025)

RAR: Retrieving And Ranking Augmented MLLMs for Visual Recognition
di: Liu, Ziyu, et al.
Pubblicazione: (2024)

Dense Connector for MLLMs
di: Yao, Huanjin, et al.
Pubblicazione: (2024)

IDPruner: Harmonizing Importance and Diversity in Visual Token Pruning for MLLMs
di: Tan, Yifan, et al.
Pubblicazione: (2026)

The Mirage of Performance Gains: Why Contrastive Decoding Fails to Mitigate Object Hallucinations in MLLMs?
di: Yin, Hao, et al.
Pubblicazione: (2025)

GranViT: A Fine-Grained Vision Model With Autoregressive Perception For MLLMs
di: Zheng, Guanghao, et al.
Pubblicazione: (2025)

Revealing Multi-View Hallucination in Large Vision-Language Models
di: Park, Wooje, et al.
Pubblicazione: (2026)

One Layer Is Enough: Adapting Pretrained Visual Encoders for Image Generation
di: Gao, Yuan, et al.
Pubblicazione: (2025)

EVE: Verifiable Self-Evolution of MLLMs via Executable Visual Transformations
di: Heng, Yongrui, et al.
Pubblicazione: (2026)