一次对齐即可解释：面向可扩展 B-cos 化的基础视觉 Transformer 特征对齐
Align Once to Explain: Feature Alignment for Scalable B-cosification of Foundational Vision Transformers
提出一种免标签的一次性特征对齐方法，将任意预训练视觉基础模型（监督、自监督、视觉语言）转换为内在可解释的 B-cos 变体，且转换后模型可直接作为骨干网络用于下游任务，为视觉模型可解释性提供高效、可扩展的解决方案。
原文依据 one-time, label-free feature alignment based approach that efficiently converts foundational vision models into inherently interpretable B-cos variants
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论基础视觉模型因其强大性能已成为许多视觉任务的事实标准。然而，它们以不透明著称，且难以解释。我们提出 ALOE（一次对齐即可解释），这是一种基于一次性、免标签特征对齐的方法，能够高效地将基础视觉模型转换为内在可解释的 B-cos 变体。一旦完成对齐，B-cos 骨干网络即可作为即插即用替代品用于多个下游任务——从而分摊可解释性的成本。ALOE 对多种预训练范式（监督、自监督、视觉语言）均具有鲁棒性，且比从头训练的数据效率高 100-1000 倍。在分类任务上，它优于全监督 B-cos 模型（例如，对于 ViT-B/16，在 ImageNet 上 top-1 准确率提升 9.2 个百分点），并保持强大的线性探测、k-NN 和零样本迁移性能，在不同数据集上与基础骨干网络（DINOv3、SigLIP2）相比具有竞争力。它还保留了对于密集预测有用的空间结构化特征，同时通过设计产生定位良好且高度人类可解释的解释。代码链接：https://github.com/rmaser/aloe。
英文原文摘要 1148 字符
Foundational vision models have become the de facto standard for many vision tasks due to their strong performance. However, they are notoriously opaque and remain hard to interpret. We present ALOE (ALign Once to Explain), a one-time, label-free feature alignment based approach that efficiently converts foundational vision models into inherently interpretable B-cos variants. Once aligned, the B-cos backbone is used as a drop-in replacement across several downstream tasks--amortizing the cost of interpretability. ALOE is robust across pre-training paradigms (supervised, self-supervised, vision-language) and is 100-1000x more data-efficient than training from scratch. On classification, it outperforms fully-supervised B-cos models (e.g., +9.2 p.p. top-1 on ImageNet for ViT-B/16), retains strong linear probing, k-NN, and zero-shot transfer performance competitive with foundational backbones (DINOv3, SigLIP2) across diverse datasets. It also preserves spatially structured features useful for dense prediction, while yielding well-localized and highly human-interpretable explanations by design. Code link https://github.com/rmaser/aloe.