Who Brings the Frisbee: Probing Hidden Hallucination Factors in Large Vision-Language Model via Causality Analysis
This addresses the issue of hallucination eroding user trust in LVLMs for real-world applications, but it appears incremental as it builds on existing understanding of the problem.
The study tackled the problem of hallucination in large vision-language models, where they generate non-existent visual elements, by proposing a causal approach to identify hidden factors like objects and contexts; the result was a technique that significantly reduces hallucinations, though no concrete numbers are provided.
Recent advancements in large vision-language models (LVLM) have significantly enhanced their ability to comprehend visual inputs alongside natural language. However, a major challenge in their real-world application is hallucination, where LVLMs generate non-existent visual elements, eroding user trust. The underlying mechanism driving this multimodal hallucination is poorly understood. Minimal research has illuminated whether contexts such as sky, tree, or grass field involve the LVLM in hallucinating a frisbee. We hypothesize that hidden factors, such as objects, contexts, and semantic foreground-background structures, induce hallucination. This study proposes a novel causal approach: a hallucination probing system to identify these hidden factors. By analyzing the causality between images, text prompts, and network saliency, we systematically explore interventions to block these factors. Our experimental findings show that a straightforward technique based on our analysis can significantly reduce hallucinations. Additionally, our analyses indicate the potential to edit network internals to minimize hallucinated outputs.