用合成图像测试概念模型在分布偏移下的泛化能力
SUB: Benchmarking CBM Generalization via Synthetic Attribute Substitutions
- 构建38,400张合成鸟图，替换特定属性如翼色、腹纹
- 提出共享噪声的扩散引导方法，确保图像属性与类别一致
- 为可解释模型提供细粒度评估基准，适合医疗等高风险领域研究者
概念瓶颈模型（CBMs）等基于概念的可解释模型在提升AI透明度方面前景广阔，尤其在医学等关键领域。然而我们发现，当数据分布发生偏移时，CBMs难以可靠识别正确概念。为此，我们引入SUB：一个基于CUB数据集的细粒度图像与概念基准，包含38,400张合成图像。通过选取33个鸟类类别和45个概念，生成替换特定属性（如翼色或腹纹）的图像。我们提出一种新型纠缠扩散引导（Tied Diffusion Guidance, TDG）方法，通过在两个并行去噪过程中共享噪声，确保生成图像既符合正确的鸟种又具备正确属性。该基准支持对CBMs及类似可解释模型进行严格评估，推动更鲁棒方法的发展。代码与数据集已开源。
原文摘要 · Abstract (English)
Concept Bottleneck Models (CBMs) and other concept-based interpretable models show great promise for making AI applications more transparent, which is essential in fields like medicine. Despite their success, we demonstrate that CBMs struggle to reliably identify the correct concepts under distribution shifts. To assess the robustness of CBMs to concept variations, we introduce SUB: a fine-grained image and concept benchmark containing 38,400 synthetic images based on the CUB dataset. To create SUB, we select a CUB subset of 33 bird classes and 45 concepts to generate images which substitute a specific concept, such as wing color or belly pattern. We introduce a novel Tied Diffusion Guidance (TDG) method to precisely control generated images, where noise sharing for two parallel denoising processes ensures that both the correct bird class and the correct attribute are generated. This novel benchmark enables rigorous evaluation of CBMs and similar interpretable models, contributing to the development of more robust methods. Our code is available at https://github.com/ExplainableML/sub and the dataset at http://huggingface.co/datasets/Jessica-bader/SUB.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。