空间基础的概念图像分类
Spatially Grounded Concept-Based Image Classification
AI总结：
提出SEG-MIL-CBM，通过概念引导区域和注意力聚合实现空间基础的概念瓶颈模型，提升最差组准确率并保持标准识别竞争力。
摘要：
深度神经网络可以在依赖难以检查或与预期任务不一致的证据时达到高精度。概念瓶颈模型（CBM）暴露了人类可解释的概念，但大多数将概念视为全局属性，并未展示局部证据如何聚合为决策。我们提出 \textbf{SEG-MIL-CBM}，一种空间基础的CBM，它将每个图像分解为概念引导的区域，并通过基于注意力的段级概念证据聚合进行分类。相同的段证据项构成预测和解释，无需单独的事后归因模块即可揭示哪些区域和概念支持预测的logit。在评估的CBM系列基线中，SEG-MIL-CBM将Waterbirds最差组准确率从$65.1\%$提高到$72.0\%$，在Pawrious上达到$87.4\%$的最差组准确率，在标准识别上保持竞争力，并在CIFAR-100上达到最佳的CBM准确率（$85.3\%$）。在CUB上的段级忠实性实验进一步表明，其学习的段排名匹配或优于评估的段排名控制。
英文摘要：
Deep neural networks can achieve high accuracy while relying on evidence that is hard to inspect or misaligned with the intended task. Concept Bottleneck Models (CBMs) expose human-interpretable concepts, but most treat concepts as global attributes and do not show how localized evidence is aggregated into a decision. We propose \textbf{SEG-MIL-CBM}, a spatially grounded CBM that decomposes each image into concept-guided regions and classifies it by attention-based aggregation of segment-level concept evidence. The same segment evidence terms form the prediction and the explanation, exposing which regions and concepts support the predicted logit without a separate post-hoc attribution module. Among evaluated CBM-family baselines, SEG-MIL-CBM improves Waterbirds worst-group accuracy from $65.1\%$ to $72.0\%$, reaches $87.4\%$ worst-group accuracy on Pawrious, remains competitive on standard recognition, and attains the best CBM accuracy on CIFAR-100 ($85.3\%$). Segment-level faithfulness experiments on CUB further show that its learned segment ranking matches or improves over evaluated segment-ranking controls.
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。
