Research map / Label-efficient and robust learning / Test-time adaptation
Cloud-Device Collaborative Adaptation to Continual Changing Environments in the Real-World
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use U-VPA cloud-device adaptation instead of source-only inference | FasterRCNN ResNet-18 · Cityscapes-to-Cityscapes-C | mAP@0.5 (%): 17.4 -> 31.0 | Helped |
| Use U-VPA cloud-device adaptation instead of source-only inference | FasterRCNN ResNet-18 · Cityscapes-to-ACDC-Detection | mAP@0.5 (%): 14.1 -> 18.5 | Helped |
| Add VPA to pseudo-label training (ablation) | FasterRCNN ResNet-18 on Cityscapes-to-Cityscapes-C · Cityscapes-to-Cityscapes-C | mAP (%): 24.1 -> 27.4 | Helped |
| Add UGS to pseudo-label training with VPA (ablation) | FasterRCNN ResNet-18 on Cityscapes-to-Cityscapes-C · Cityscapes-to-Cityscapes-C | mAP (%): 27.4 -> 28.3 | Helped |
| Add U-EMA to pseudo-label training with VPA and UGS (ablation) | FasterRCNN ResNet-18 on Cityscapes-to-Cityscapes-C · Cityscapes-to-Cityscapes-C | mAP (%): 28.3 -> 31.0 | Helped |
| Select samples by uncertainty rather than confidence (ablation) | U-VPA · Cityscapes-to-Cityscapes-C | mAP (%): 27.4 -> 31.0 | Helped |
| Use UGS rather than random selection for 25% of the data (ablation) | Cloud-device adaptation · Cityscapes-to-Cityscapes-C | mAP (%): 23.5 -> 28.2 | Helped |
| Use UGS with 50% of data rather than random selection with all data (ablation) | Cloud-device adaptation · Cityscapes-to-Cityscapes-C | mAP (%): 28.3 -> 31.0 | Helped |
| Use UGS with 75% of data rather than random selection with all data (ablation) | Cloud-device adaptation · Cityscapes-to-Cityscapes-C | mAP (%): 28.3 -> 31.2 | Helped |
Closest papers by meaning
- Cloud-Device Collaborative Learning for Multimodal Large Language Models (CVPR 2024 · similarity 0.88)
- PCoTTA: Continual Test-Time Adaptation for Multi-Task Point Cloud Understanding (NeurIPS 2024 · similarity 0.84)
- Cross-Device Collaborative Test-Time Adaptation (NeurIPS 2024 · similarity 0.83)
- Dynamic Adapter Meets Prompt Tuning: Parameter-Efficient Transfer Learning for Point Cloud Analysis (CVPR 2024 · similarity 0.82)
- Towards Robust and Efficient Cloud-Edge Elastic Model Adaptation via Selective Entropy Distillation (ICLR 2024 · similarity 0.82)
- Bridging On-Device and Cloud LLMs for Collaborative Reasoning: A Unified Methodology for Local Routing and Post-Training (ICML 2026 · similarity 0.82)
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.