CoFiDA-M：面向跨域自适应且仅需图像推理的概念感知特征调制
CoFiDA-M: Concept-Aware Feature Modulation for Cross-Domain Adaptation with Image-Only Inference
把只在训练期可得的稠密概率元数据（MONET 概念分数）经 FiLM 调制教师特征，再让学生回归被『编辑』后的特征空间而非仅拟合 logits，这一特权信息蒸馏范式可迁移到任何测试时缺失辅助模态的视觉任务。
原文依据 trains a teacher network that uses MONET concept probabilities to guide a FiLM modulator
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于人工智能的皮肤癌筛查模型在从专家级皮肤镜（源域）图像迁移到消费级临床（目标域）照片时性能会严重下降，阻碍了其实际部署。现有的域自适应方法往往忽略关键的语义不变量，例如临床概念。尽管 MONET 等新型基础模型能够以稠密概率分数的形式提供这类语义信息，但该元数据在测试时不可获得，这为实用的仅图像筛查工具造成了部署悖论。为填补这一空白，我们提出 CoFiDA-M，一个特权信息框架，在训练时从概念中学习，但部署时作为仅图像模型运行。我们的方法训练一个教师网络，利用 MONET 概念概率引导 FiLM 调制器，将视觉特征变换到语义上被“编辑”过的特征空间。随后训练一个轻量级的仅图像学生网络来复现这一被编辑的表征，而不仅仅是复现教师的最终预测。这种蒸馏将临床推理“烘焙”进学生的权重中。在具有挑战性的多数据集基准上，我们的仅图像学生模型显著优于最优方法，尤其在黑色素瘤召回率上表现突出。我们的工作提供了一个实用且可泛化的框架，可将含噪的概率元数据用作特权信息，展现出强大的跨数据集鲁棒性以及在皮肤科之外实际部署的潜力。代码将在论文录用后公开。
英文原文摘要 1518 字符
Models for AI-based skin cancer screening suffer a severe performance drop when shifting from expert dermoscopic (source) images to consumer-grade clinical (target) photos, hindering real-world deployment. Existing domain adaptation methods often ignore crucial semantic invariants, such as clinical concepts. While new foundation models like MONET can provide this semantic information as dense, probabilistic scores, this metadata is unavailable at test time, creating a deployment paradox for practical image-only screening tools. We address this gap by proposing CoFiDA-M, a privileged information framework that learns from concepts at training time but deploys as an image-only model. Our method trains a teacher network that uses MONET concept probabilities to guide a FiLM modulator, transforming visual features into a semantically "edited" feature space. A lightweight, image-only student is then trained to reproduce this edited representation, not just the teacher's final predictions. This distillation "bakes" the clinical reasoning into the student's weights.On a challenging multi-dataset benchmark, our image-only student significantly outperforms state-of-the-art approaches, especially in melanoma recall. Our work provides a practical and generalizable framework for leveraging noisy, probabilistic metadata as privileged information, demonstrating strong cross-dataset robustness and potential for real-world deployment beyond dermatology. The code will be made publicly available upon acceptance.