Imagen de Portada

MHSA: A Lightweight Framework for Mitigating Hallucinations via Steered Attention in LVLMs

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Ding, Wei, Li, Yilin, Zhang, Yudong, Xie, Ruobing, Sun, Xingwu, Chen, Jiansheng, Wang, Yu
Formato:	Preprint
Publicado:	2026
Materias:	Computer Vision and Pattern Recognition Artificial Intelligence
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

PIP: Detecting Adversarial Examples in Large Vision-Language Models via Attention Patterns of Irrelevant Probe Questions
por: Zhang, Yudong, et al.
Publicado: (2024)

DHCP: Detecting Hallucinations by Cross-modal Attention Pattern in Large Vision-Language Models
por: Zhang, Yudong, et al.
Publicado: (2024)

Fighting Fire with Fire (F3): A Training-free and Efficient Visual Adversarial Example Purification Method in LVLMs
por: Zhang, Yudong, et al.
Publicado: (2025)

Enhancing Contrastive Learning Inspired by the Philosophy of "The Blind Men and the Elephant"
por: Zhang, Yudong, et al.
Publicado: (2024)

QAVA: Query-Agnostic Visual Attack to Large Vision-Language Models
por: Zhang, Yudong, et al.
Publicado: (2025)

Mitigating Object Hallucinations in LVLMs via Attention Imbalance Rectification
por: Sun, Han, et al.
Publicado: (2026)

Mitigating Hallucination in Multimodal Large Language Model via Hallucination-targeted Direct Preference Optimization
por: Fu, Yuhan, et al.
Publicado: (2024)

Steering LVLMs via Sparse Autoencoder for Hallucination Mitigation
por: Hua, Zhenglin, et al.
Publicado: (2025)

TARAC: Mitigating Hallucination in LVLMs via Temporal Attention Real-time Accumulative Connection
por: Jiang, Lei, et al.
Publicado: (2025)

Causally-Grounded Dual-Path Attention Intervention for Object Hallucination Mitigation in LVLMs
por: Yu, Liu, et al.
Publicado: (2025)

Optimizing LVLMs with On-Policy Data for Effective Hallucination Mitigation
por: Yu, Chengzhi, et al.
Publicado: (2025)

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

Attention-space Contrastive Guidance for Efficient Hallucination Mitigation in LVLMs
por: Jo, Yujin, et al.
Publicado: (2026)

Countering the Over-Reliance Trap: Mitigating Object Hallucination for LVLMs via a Self-Validation Framework
por: Liu, Shiyu, et al.
Publicado: (2026)

Magnifier Prompt: Tackling Multimodal Hallucination via Extremely Simple Instructions
por: Fu, Yuhan, et al.
Publicado: (2024)

Identify, Isolate, and Purge: Mitigating Hallucinations in LVLMs via Self-Evolving Distillation
por: Li, Wenhao, et al.
Publicado: (2025)

Seeing Clearly by Layer Two: Enhancing Attention Heads to Alleviate Hallucination in LVLMs
por: Zhang, Xiaofeng, et al.
Publicado: (2024)

MRFD: Multi-Region Fusion Decoding with Self-Consistency for Mitigating Hallucinations in LVLMs
por: Ge, Haonan, et al.
Publicado: (2025)

CATCH: Complementary Adaptive Token-level Contrastive Decoding to Mitigate Hallucinations in LVLMs
por: Kan, Zhehan, et al.
Publicado: (2024)

Mitigating Hallucinations in Large Vision-Language Models (LVLMs) via Language-Contrastive Decoding (LCD)
por: Manevich, Avshalom, et al.
Publicado: (2024)

Seeing It or Not? Interpretable Vision-aware Latent Steering to Mitigate Object Hallucinations
por: Chen, Boxu, et al.
Publicado: (2025)

Dynamic Multimodal Activation Steering for Hallucination Mitigation in Large Vision-Language Models
por: Yin, Jianghao, et al.
Publicado: (2026)

Mitigating Entangled Steering in Large Vision-Language Models for Hallucination Reduction
por: Zhang, Yuanhong, et al.
Publicado: (2026)

What Makes LVLMs Hallucinate Less? Unveiling the Architectural Factors Behind Hallucination Robustness
por: He, Yusheng, et al.
Publicado: (2026)

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

Towards Interpretable Hallucination Analysis and Mitigation in LVLMs via Contrastive Neuron Steering
por: Lyu, Guangtao, et al.
Publicado: (2026)

CHASD: Language Increment-Calibrated Contrastive Decoding against Hallucination in LVLMs
por: Huang, Xiaoyi, et al.
Publicado: (2026)

PhD: A ChatGPT-Prompted Visual hallucination Evaluation Dataset
por: Liu, Jiazhen, et al.
Publicado: (2024)

Mitigating Object Hallucinations in Large Vision-Language Models via Attention Calibration
por: Zhu, Younan, et al.
Publicado: (2025)

Cross-Modal Attention Calibration for LVLM Hallucination Mitigation
por: Li, Jiaming, et al.
Publicado: (2025)

GLSim: Detecting Object Hallucinations in LVLMs via Global-Local Similarity
por: Park, Seongheon, et al.
Publicado: (2025)

When RAG Hurts: Diagnosing and Mitigating Attention Distraction in Retrieval-Augmented LVLMs
por: Zhao, Beidi, et al.
Publicado: (2026)

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

Attention Hijackers: Detect and Disentangle Attention Hijacking in LVLMs for Hallucination Mitigation
por: Chen, Beitao, et al.
Publicado: (2025)

Adaptive Residual-Update Steering for Low-Overhead Hallucination Mitigation in Large Vision Language Models
por: Zou, Zhengtao, et al.
Publicado: (2025)

Mitigating Hallucination in Vision-Language Models through Barrier-Regulated Adaptive Closed-form Steering
por: Jana, Soumyadeep, et al.
Publicado: (2026)

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

Mitigating Cross-Image Information Leakage in LVLMs for Multi-Image Tasks
por: Park, Yeji, et al.
Publicado: (2025)

VidLBEval: Benchmarking and Mitigating Language Bias in Video-Involved LVLMs
por: Yang, Yiming, et al.
Publicado: (2025)

Visual Description Grounding Reduces Hallucinations and Boosts Reasoning in LVLMs
por: Ghosh, Sreyan, et al.
Publicado: (2024)