Research map / Continual learning / Domain and prompt-based
Cross-Architecture Adaptation: Cloud-Edge Continual Test-Time Adaptation with Dynamic Sampling and Heterogeneous Distillation
Zirui Xu, Xianhang Chu, Jiahao Li, Xu Yang, Cheng Deng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Multi-level Adaptive Heterogeneous Distillation (MAHD) instead of plain cross-entropy distillation (ablation) | CAA edge ResNet-18 · ImageNet-C severity 5 | average accuracy (%): 36.2 -> 41.2 | Improved |
| add cross-domain detection and reset mechanism (ablation) | CAA · ImageNet-C severity 5 | average accuracy (%): 40.3 -> 41.2 | Improved |
| use heterogeneous ViT-B cloud teacher with MAHD for ResNet-18 edge instead of CoLA | edge ResNet-18 CTTA · ImageNet-C severity 5 | average accuracy (%): 38.6 -> 41.2 | Improved |
| use MDCS uncertainty-representativeness sampling instead of baseline uploading | CAA · ImageNet-C severity 5 | uploaded samples (K): 369.0K (CEMA) -> 335.1K | Improved |
| CAA vs CEMA on Contrast corruption | CAA · ImageNet-C severity 5, Contrast | accuracy (%): 35.4 (CEMA) -> 33.0 | Hurt |
| CAA vs ETA/CoLA on Defocus blur | CAA · ImageNet-C severity 5, Defocus | accuracy (%): 26.3 (CoLA) -> 26.2 | No clear effect |
| apply non-adaptive or output-only adaptation (LAME) to edge ResNet-18 | ResNet-18 · ImageNet-C severity 5 | average accuracy (%): 14.7 -> 13.9 | No clear effect |
Closest papers by meaning
- Towards Robust and Efficient Cloud-Edge Elastic Model Adaptation via Selective Entropy Distillation (ICLR 2024 · similarity 0.90)
- SNAP: Low-Latency Test-Time Adaptation with Sparse Updates (NeurIPS 2025 · similarity 0.87)
- AdMiT: Adaptive Multi-Source Tuning in Dynamic Environments (CVPR 2025 · similarity 0.86)
- BECoTTA: Input-dependent Online Blending of Experts for Continual Test-time Adaptation (ICML 2024 · similarity 0.85)
- HybridFlow: Resource-Adaptive Subtask Routing for Efficient Edge-Cloud LLM Inference (ICML 2026 · similarity 0.85)
- TinyTTA: Efficient Test-time Adaptation via Early-exit Ensembles on Edge Devices (NeurIPS 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
