Research map / Label-efficient and robust learning / Test-time adaptation
Dance Across Shifts: Forward-Facilitation Continual Test-Time Adaptation through Dynamic Style Bridging
Zhilin Zhu, Yabin Wang, Zhiheng Ma, Yaguang Song, Yaowei Wang, Xiaopeng Hong
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 synthetic knowledge base with semantic loss (ablation) | Self-training baseline · ImageNet-to-ImageNetC | mean error: 50.0 -> 47.4 | Improved |
| add input-level bridging (ablation) | Ex2 · ImageNet-to-ImageNetC | mean error: 47.4 -> 45.8 | Improved |
| add statistic-level bridging (ablation) | Ex2 · ImageNet-to-ImageNetC | mean error: 47.4 -> 45.4 | Improved |
| add representation-level bridging (L_SCL) (ablation) | Ex2 · ImageNet-to-ImageNetC | mean error: 47.4 -> 46.5 | Improved |
| combine input and statistic bridging (ablation) | Ex2 · ImageNet-to-ImageNetC | mean error: 47.4 -> 44.6 | Improved |
| add representation-level bridging on top of input and statistic (ablation) | Ex6 · ImageNet-to-ImageNetC | mean error: 44.6 -> 44.1 | Improved |
| use SD 1.5 instead of BigGAN to build knowledge base (ablation) | Ours · ImageNet-to-ImageNetC | mean error: 45.0 -> 44.1 | Improved |
| use SD 3.0 instead of SD 1.5 (ablation) | Ours · ImageNet-to-ImageNetC | mean error: 44.1 -> 43.8 | Improved |
| add synthetic knowledge bridging to entropy-minimization self-training (ablation) | Entropy minimization · ImageNet-to-ImageNetC | mean error: 49.8 -> 44.8 | Improved |
| add synthetic knowledge bridging to teacher-student self-training (ablation) | Teacher-student · ImageNet-to-ImageNetC | mean error: 50.0 -> 44.1 | Improved |
| bridging method versus SDA (diffusion denoising) | Ours · CIFAR100-to-CIFAR100C | mean error: 42.9 -> 29.8 | Improved |
| bridging method versus DPCore on ImageNetC | Ours · ImageNet-to-ImageNetC | mean error: 47.6 -> 44.1 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- PAID — PAID: Pairwise Angular-Invariant Decomposition for Continual Test-Time Adaptation (NeurIPS 2025) · better on 3 of 3 shared results
- TENT — Tent: Fully Test-Time Adaptation by Entropy Minimization (ICLR 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- DPCore: Dynamic Prompt Coreset for Continual Test-Time Adaptation (ICML 2025 · similarity 0.87)
- A Versatile Framework for Continual Test-Time Domain Adaptation: Balancing Discriminability and Generalizability (CVPR 2024 · similarity 0.86)
- ViDA: Homeostatic Visual Domain Adapter for Continual Test Time Adaptation (ICLR 2024 · similarity 0.85)
- PAID: Pairwise Angular-Invariant Decomposition for Continual Test-Time Adaptation (NeurIPS 2025 · similarity 0.85)
- ReservoirTTA: Prolonged Test-time Adaptation for Evolving and Recurring Domains (NeurIPS 2025 · similarity 0.85)
- PALM: Pushing Adaptive Learning Rate Mechanisms for Continual Test-Time Adaptation (AAAI 2025 · similarity 0.84)
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.
