13 papers · 1 filter
EviAnchor: Mitigating Hallucinations in Large Vision-Language Models via Regional Visual Evidence Compensation
Sihang Jia, Shuliang Liu, Songbo Yang +1
Large vision-language models (LVLMs) frequently generate content unsupported by visual inputs. Preliminary experiments show that visual evidence is primarily incorporated into answ…
AnchorFold: A Focus-Then-Fold Framework via Recursive Attention Propagation for Efficient Multi-Vector Visual Document Retrieval
Haoyu Zuo, Yibo Yan, Xin Zou +4
Multi-vector vision-language retrievers enable fine-grained Visual Document Retrieval (VDR) through late interaction, but storing and scoring hundreds of visual patch embeddings pe…
Clearer Sight, Fewer Lies: Oriented Pickup Preference Optimization for Multimodal Hallucination Mitigation
Xin Zou, Haolin Deng, Yibo Yan +5
Multimodal Large Language Models (MLLMs) are prone to hallucination as their generation preferences are insufficiently calibrated to visual evidence, causing them to fall back on l…
Consistency as Inductive Bias: Learning Cross-View Invariance for Robust Multimodal Reasoning
Xin Zou, Haolin Deng, Yibo Yan +6
Inductive biases steer learning toward generalizable solutions by encoding task structure. In this work, we identify a crucial missing bias in MLLMs: cross-view consistency, \texti…
Decoding by Perturbation: Mitigating MLLM Hallucinations via Dynamic Textual Perturbation
Sihang Jia, Shuliang Liu, Songbo Yang +3
Multimodal Large Language Models frequently suffer from inference hallucinations, partially stemming from language priors dominating visual evidence. Existing training-free mitigat…
Visual Late Chunking: An Empirical Study of Contextual Chunking for Efficient Visual Document Retrieval
Yibo Yan, Mingdong Ou, Yi Cao +5
Multi-vector models dominate Visual Document Retrieval (VDR) due to their fine-grained matching capabilities, but their high storage and computational costs present a major barrier…
Correct Prediction, Wrong Steps? Consensus Reasoning Knowledge Graph for Robust Chain-of-Thought Synthesis
Zipeng Ling, Shuliang Liu, Seonil Son +4
Large language models (LLMs) have become increasingly used for various tasks, often coupled with Chain-of-Thought (CoT) prompting to boost accuracy. Recent work has shown that high…
Beyond the Grid: Layout-Informed Multi-Vector Retrieval with Parsed Visual Document Representations
Yibo Yan, Mingdong Ou, Yi Cao +6
Harnessing the full potential of visually-rich documents requires retrieval systems that understand not just text, but intricate layouts, a core challenge in Visual Document Retrie…
Unlocking Multimodal Document Intelligence: From Current Triumphs to Future Frontiers of Visual Document Retrieval
Yibo Yan, Jiahao Huo, Guanbo Feng +12
With the rapid proliferation of multimodal information, Visual Document Retrieval (VDR) has emerged as a critical frontier in bridging the gap between unstructured visually rich da…
Sculpting the Vector Space: Towards Efficient Multi-Vector Visual Document Retrieval via Prune-then-Merge Framework
Yibo Yan, Mingdong Ou, Yi Cao +5
Visual Document Retrieval (VDR), which aims to retrieve relevant pages within vast corpora of visually-rich documents, is of significance in current multimodal retrieval applicatio…
A Visual Semantic Adaptive Watermark grounded by Prefix-Tuning for Large Vision-Language Model
Qi Zheng, Shuliang Liu, Yu Huang +8
Watermarking has emerged as a pivotal solution for content traceability and intellectual property protection in Large Vision-Language Models (LVLMs). However, vision-agnostic water…
Vision-Language Introspection: Mitigating Overconfident Hallucinations in MLLMs via Interpretable Bi-Causal Steering
Shuliang Liu, Songbo Yang, Dong Fang +7
Object hallucination critically undermines the reliability of Multimodal Large Language Models, often stemming from a fundamental failure in cognitive introspection, where models b…
Distilling the Thought, Watermarking the Answer: A Principle Semantic Guided Watermark for Large Reasoning Models
Shuliang Liu, Xingyu Li, Hongyi Liu +6
Reasoning Large Language Models (RLLMs) excelling in complex tasks present unique challenges for digital watermarking, as existing methods often disrupt logical coherence or incur…