保持冻结：面向多域视觉 Transformer 的域路由条件残差调制
Keep It Frozen: Domain-Routed Conditional Residual Modulation for Multi-Domain Vision Transformers
提出一种冻结骨干网络、通过域路由和参数合成网络动态生成低秩残差调制块的方法，为视觉 Transformer 在多域（医学+自然）场景下提供参数高效的自适应方案，可复用于需要跨域泛化且计算开销敏感的任务。
原文依据 DCRM-ViT keeps the backbone frozen and augments each block with lightweight Residual Modulation Blocks (RMBs), whose parameters are synthesized per sample by a Domain Router (DR) and a Parameter Synthesizer Network (PSN).
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论医学影像由于声影、运动模糊和边界模糊等问题仍然具有挑战性，而将视觉模型适应到此类域通常需要大量任务特定的微调，并且可能降低通用图像能力。我们提出 DCRM-ViT，一种面向视觉 Transformer 的域条件残差调制框架，在适应多种医学和自然域的同时保留通用视觉知识。DCRM-ViT 保持骨干网络冻结，并在每个块中增加轻量级残差调制块（RMB），其参数由域路由器（DR）和参数合成网络（PSN）逐样本合成。DR 从输入特征预测软域权重，PSN 将其映射为低秩残差，以调制选定的投影，并可选地添加域感知注意力偏置。我们采用双层优化方案训练模型，其中内循环使 RMB 适应任务监督，外循环更新 DR、PSN 和 RMB 初始化以提升跨域泛化能力。在细粒度分类（Food101、SUN397、Stanford Cars）和医学分割（超声、CT、MRI）任务上，DCRM-ViT 以适中的可训练计算量持续优于强基线。消融研究证实了所提出组件的贡献。总体而言，DCRM-ViT 以低开销实现了强大的跨域性能，仅需 4.7 GFLOPs 和 0.3 分钟/epoch。
英文原文摘要 1430 字符
Medical imaging remains challenging due to acoustic shadows, motion blur, and indistinct boundaries, while adapting vision models to such domains often requires heavy task-specific fine-tuning and can degrade general-image capability. We propose DCRM-ViT, a domain-conditioned residual modulation framework for Vision Transformers that preserves general-vision knowledge while adapting to diverse medical and natural domains. DCRM-ViT keeps the backbone frozen and augments each block with lightweight Residual Modulation Blocks (RMBs), whose parameters are synthesized per sample by a Domain Router (DR) and a Parameter Synthesizer Network (PSN). The DR predicts soft domain weights from input features, and the PSN maps them to low-rank residuals that modulate selected projections and optionally add a domain-aware attention bias. We train the model with a bi-level optimization scheme, where an inner loop adapts RMBs to task supervision and an outer loop updates DR, PSN, and RMB initialization to improve cross-domain generalization. Across fine-grained classification (Food101, SUN397, Stanford Cars) and medical segmentation (ultrasound, CT, MRI), DCRM-ViT consistently outperforms strong baselines with modest trainable compute. Ablation studies confirm the contribution of the proposed components. Overall, DCRM-ViT achieves strong cross-domain performance with low overhead, requiring only 4.7 GFLOPs and 0.3 min/epoch.