视觉概念排名揭示大型多模态模型使用的医学捷径
Visual concept ranking uncovers medical shortcuts used by large multimodal models
AI总结：
本文提出视觉概念排名方法，用于揭示大型多模态模型在医疗任务中表现的潜在视觉特征依赖性。
摘要：
确保在安全关键领域如医疗保健中机器学习模型的可靠性需要能够揭示模型缺陷的审计方法。我们介绍了一种方法，用于识别大型多模态模型（LMMs）中的重要视觉概念，并利用该方法来研究这些模型在被提示进行医疗任务时表现出的行为。我们主要关注从临床皮肤病学图像中分类恶性皮肤病变的任务，补充实验包括胸部X光和自然图像。在展示LMMs在被提示示例时不同人口子群体之间表现出现意外差距后，我们应用我们的方法，视觉概念排名（VCR），到这些模型和提示中。VCR生成与不同视觉特征依赖性相关的假设，我们随后能够通过手动干预验证这些假设。
英文摘要：
Ensuring the reliability of machine learning models in safety-critical domains such as healthcare requires auditing methods that can uncover model shortcomings. We introduce a method for identifying important visual concepts within large multimodal models (LMMs) and use it to investigate the behaviors these models exhibit when prompted with medical tasks. We primarily focus on the task of classifying malignant skin lesions from clinical dermatology images, with supplemental experiments including both chest radiographs and natural images. After showing how LMMs display unexpected gaps in performance between different demographic subgroups when prompted with demonstrating examples, we apply our method, Visual Concept Ranking (VCR), to these models and prompts. VCR generates hypotheses related to different visual feature dependencies, which we are then able to validate with manual interventions.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
