Research map / Label-efficient and robust learning / Test-time adaptation
Decoupling Stability and Plasticity for Multi-Modal Test-Time Adaptation
Yongbo He, Zirun Guo, Tao Jin
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 stable adapter (ablation) | DASP · Kinetics50-C (avg of Video-C, Audio-C, Interleaved-C) | accuracy (%): 67.4 -> 68.4 | Improved |
| add plastic adapter (ablation) | DASP · Kinetics50-C (avg of Video-C, Audio-C, Interleaved-C) | accuracy (%): 66.0 -> 68.4 | Improved |
| add asymmetric adaptation (ablation) | DASP · Kinetics50-C (avg of Video-C, Audio-C, Interleaved-C) | accuracy (%): 62.5 -> 68.4 | Improved |
| use asymmetric adaptation as designed instead of opposite configuration (ablation) | DASP · Kinetics50-C (avg of Video-C, Audio-C, Interleaved-C) | accuracy (%): 61.9 -> 68.4 | Improved |
| add asymmetric adaptation (ablation) | DASP · Kinetics50-C Video-C | accuracy (%): 54.3 -> 65.3 | Improved |
| add stable adapter (ablation) | DASP · Kinetics50-C Interleaved-C | accuracy (%): 65.2 -> 67.3 | Improved |
| add stable adapter (ablation) | DASP · Kinetics50-C Audio-C | accuracy (%): 72.3 -> 72.5 | No clear effect |
| use DASP over TSA under audio corruption | TSA · VGGSound-C audio (episodic) | accuracy (%): 34.7 -> 39.7 | Improved |
| use DASP over TSA under interleaved corruption | TSA · Kinetics50-C interleaved (continual) | accuracy (%): 63.9 -> 67.3 | Improved |
| use DASP instead of TSA on VGGSound-C interleaved continual | TSA · VGGSound-C interleaved (continual) | accuracy (%): 12.8 -> 52.5 | Improved |
| use DASP instead of Tent in continual video adaptation | Tent · Kinetics50-C video (continual) | accuracy (%): 41.8 -> 65.3 | Improved |
| use DASP over TSA on VGGSound-C video corruption | TSA · VGGSound-C video (episodic) | accuracy (%): 56.6 -> 57.5 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most 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
- Test-Time Selective Adaptation for Uni-Modal Distribution Shift in Multi-Modal Data (ICML 2025 · similarity 0.91)
- Attention Bootstrapping for Multi-Modal Test-Time Adaptation (AAAI 2025 · similarity 0.89)
- Towards Robust Multimodal Open-set Test-time Adaptation via Adaptive Entropy-aware Optimization (ICLR 2025 · similarity 0.88)
- Smoothing the Shift: Towards Stable Test-Time Adaptation under Complex Multimodal Noises (ICLR 2025 · similarity 0.87)
- ReservoirTTA: Prolonged Test-time Adaptation for Evolving and Recurring Domains (NeurIPS 2025 · similarity 0.87)
- Dual Prototype Evolving for Test-Time Generalization of Vision-Language Models (NeurIPS 2024 · similarity 0.87)
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.
