Interpreting and Mitigating Hallucination in MLLMs through Multi-agent Debate
This addresses a critical reliability issue in MLLMs for applications like image captioning and visual QA, though it is incremental as it builds on existing self-reflection and debate methods.
The paper tackles the problem of hallucination in multimodal large language models (MLLMs), where outputs are inconsistent with visual content, by proposing a multi-agent debate approach to mitigate hallucinations and interpret their causes, achieving generalized performance across several MLLMs in experiments.
MLLMs often generate outputs that are inconsistent with the visual content, a challenge known as hallucination. Previous methods focus on determining whether a generated output is hallucinated, without identifying which image region leads to the hallucination or interpreting why such hallucinations occur. In this paper, we argue that hallucination in MLLMs is partially due to a lack of slow-thinking and divergent-thinking in these models. To address this, we propose adopting a self-reflection scheme to promote slow-thinking. Furthermore, we consider eliminating hallucination as a complex reasoning task and propose a multi-agent debate approach to encourage divergent-thinking. Consequently, our approach can not only mitigate hallucinations but also interpret why they occur and detail the specifics of hallucination. In addition, we propose to distinguish creativity from hallucination in the context of MLLMs, and illustrate how to evaluate MLLMs' creativity capability. Extensive experiments on various benchmarks demonstrate that our approach exhibits generalized hallucinations-mitigating performance across several MLLMs.