跨架构适配：面向云边持续测试时自适应的动态采样与异构蒸馏
Cross-Architecture Adaptation: Cloud-Edge Continual Test-Time Adaptation with Dynamic Sampling and Heterogeneous Distillation
把云端大 Transformer 教师与边缘轻量 CNN 学生的异构蒸馏用于持续测试时自适应，为视觉模型在分布漂移下的在线部署提供了可复用的跨架构知识迁移与样本筛选方案。
原文依据 a Multi-level Adaptive Heterogeneous Distillation module is proposed to facilitate effective knowledge transfer across the architecturally disparate models
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论云边持续测试时自适应（CTTA）——由边缘设备处理实时数据、云端提供强大算力——是模型在真实场景中适应动态数据分布的关键范式。然而，现有框架大多假设云端与边缘 CNN 之间架构同构，这构成了显著的性能瓶颈，尤其在基于 Transformer 的模型快速涌现的背景下。现有方法未能弥合这一架构鸿沟，导致在自适应精度与实际适用性上存在明显不足。为此，我们提出一种新颖的跨架构适配（CAA）框架，用于异构云边 CTTA，能够有效适应不断变化的数据分布。具体而言，CAA 在云端部署大型 Transformer 教师模型以进行稳健的特征提取与预测，在边缘设备部署轻量 CNN 学生模型以契合资源约束。基于这样的云边模型，一种协同的边缘到云端通信策略——多准则动态跨域采样——确保仅上传最具信息量且类别均衡的样本，在最小化通信开销的同时保证稳定、无偏的自适应。此外，我们提出多级自适应异构蒸馏模块，以促进架构差异显著的模型之间进行有效的知识迁移，并提升边缘模型的学习效率。在多个基准上的实验表明，CAA 在边缘资源消耗低、边缘到云端通信开销极小的条件下取得了最优性能。
英文原文摘要 1673 字符
Cloud-Edge Continual Test-Time Adaptation (CTTA)--with edge devices processing real-time data and the cloud offering strong computing power--is a critical paradigm for models that adapt to dynamic data distributions in real-world scenarios. However, most existing frameworks assume architectural homogeneity between cloud and edge CNNs, which poses a significant performance bottleneck, particularly given the rapid emergence of Transformer-based models. Current methods fail to bridge this architectural gap, resulting in significant deficiencies in adaptation accuracy and practical applicability. To address this, we propose a novel Cross-Architecture Adaptation (CAA) framework for heterogeneous Cloud-Edge CTTA that enables effective adaptation to shifting data distributions. Specifically, CAA deploys a large Transformer-based teacher model on the cloud for robust feature extraction and prediction, and a lightweight CNN-based student model on edge devices to fit resource constraints. Based on such cloud-edge models, a synergistic edge-to-cloud communication strategy, Multi-criteria Dynamic Cross-domain Sampling, ensures only the most informative, class-balanced samples are uploaded, minimizing communication costs while guaranteeing stable, unbiased adaptation. Moreover, a Multi-level Adaptive Heterogeneous Distillation module is proposed to facilitate effective knowledge transfer across the architecturally disparate models, and improve the learning efficiency of the edge one. Experiments on several benchmarks demonstrate that CAA achieves state-of-the-art performance with low edge resource consumption and minimal edge-to-cloud communication overhead.