重新审视概念瓶颈模型：从缺陷到解决方案
Rethinking Concept Bottleneck Models: From Pitfalls to Solutions
为基于概念的预测提供可复用的评估与改进框架：用熵指标预评估概念集与数据集的匹配度，在概念激活与分类器之间插入非线性层以消除线性绕过问题，并用线性教师探针引导的蒸馏损失缩小与黑箱模型的精度差距；同时系统分析不同视觉编码器与 VLM 对精度和可解释性的影响，可直接用于构建可解释的图像分类流程。
原文依据 we propose an entropy-based metric to quantify the intrinsic suitability of a concept set for a given dataset
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论概念瓶颈模型（CBM）将预测建立在人类可理解的概念之上，但面临根本性局限：缺乏可在事前评估概念相关性的指标；存在“线性问题”，使近期的 CBM 完全绕过概念瓶颈；与不透明模型相比存在精度差距；并且缺乏对不同视觉骨干网络与视觉语言模型（VLM）影响的系统研究。我们提出 CBM-Suite，一个系统性地应对上述挑战的方法框架。首先，我们提出一种基于熵的指标，用于量化概念集对给定数据集的内在适配程度。其次，我们通过在概念激活与分类器之间插入一个非线性层来解决线性问题，从而确保模型精度能够忠实反映概念相关性。第三，我们利用由线性教师探针引导的蒸馏损失来缩小精度差距。最后，我们全面分析了不同视觉编码器、视觉语言模型与概念集如何相互作用，进而影响 CBM 的精度与可解释性。大量评估表明，CBM-Suite 能产生更准确的模型，并为改进基于概念的可解释性提供了洞见。
英文原文摘要 1230 字符
Concept Bottleneck Models (CBMs) ground predictions in human-understandable concepts but face fundamental limitations: the absence of a metric to pre-evaluate concept relevance, the "linearity problem" causing recent CBMs to bypass the concept bottleneck entirely, an accuracy gap compared to opaque models, and finally the lack of systematic study on the impact of different visual backbones and VLMs. We introduce CBM-Suite, a methodological framework to systematically addresses these challenges. First, we propose an entropy-based metric to quantify the intrinsic suitability of a concept set for a given dataset. Second, we resolve the linearity problem by inserting a non-linear layer between concept activations and the classifier, which ensures that model accuracy faithfully reflects concept relevance. Third, we narrow the accuracy gap by leveraging a distillation loss guided by a linear teacher probe. Finally, we provide comprehensive analyses on how different vision encoders, vision-language models, and concept sets interact to influence accuracy and interpretability in CBMs. Extensive evaluations show that CBM-Suite yields more accurate models and provides insights for improving concept-based interpretability.