Research map / Label-efficient and robust learning / Test-time adaptation
Multi-modal Test-time Adaptation via Adaptive Probabilistic Gaussian Calibration
Jinglin Xu, Yi Li, Chuxiong Sun, Xiao Xu, Jiangmeng Li, Fanjiang Xu
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 asymmetry rectification (AR) alone (ablation) | AdaPGC base · Kinetics50-C (video), average accuracy | accuracy: 63.81 -> 65.09 | Improved |
| add fused logits (FL) alone (ablation) | AdaPGC base · Kinetics50-C (video), average accuracy | accuracy: 63.81 -> 63.57 | No clear effect |
| add prediction alignment (PA) alone (ablation) | AdaPGC base · Kinetics50-C (video), average accuracy | accuracy: 63.81 -> 63.94 | No clear effect |
| add FL and PA together (canonical-GDA-like setting) (ablation) | AdaPGC base · Kinetics50-C (video), average accuracy | accuracy: 63.81 -> 64.9 | Improved |
| add asymmetry rectification on top of FL and PA (ablation) | AdaPGC (FL+PA) · Kinetics50-C (video), average accuracy | accuracy: 64.9 -> 66.08 | Improved |
| add all three components FL, PA, AR (ablation) | AdaPGC base · Kinetics50-C (video), average accuracy | accuracy: 63.81 -> 66.08 | Improved |
| add FL alone (ablation) | AdaPGC base · Kinetics50-C (audio), average accuracy | accuracy: 71.88 -> 72.27 | Improved |
| add AR alone (ablation) | AdaPGC base · Kinetics50-C (audio), average accuracy | accuracy: 71.88 -> 72.25 | Improved |
| add AR on top of FL and PA (ablation) | AdaPGC (FL+PA) · Kinetics50-C (audio), average accuracy | accuracy: 72.90 -> 73.19 | Improved |
| add all three components (ablation) | AdaPGC base · Kinetics50-C (audio), average accuracy | accuracy: 71.88 -> 73.19 | Improved |
| vary lambda, w_c and w_g (ablation) | AdaPGC · Kinetics50-C | accuracy | No clear effect |
| apply Tent test-time adaptation | CAV-MAE source · VGGSound-C (corrupted audio) | average accuracy: 25.6 -> 4.5 | Hurt |
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 4 of 4 shared results
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)
- Backpropagation-Free Test-Time Adaptation via Probabilistic Gaussian Alignment (NeurIPS 2025 · similarity 0.89)
- Attention Bootstrapping for Multi-Modal Test-Time Adaptation (AAAI 2025 · similarity 0.88)
- Smoothing the Shift: Towards Stable Test-Time Adaptation under Complex Multimodal Noises (ICLR 2025 · similarity 0.87)
- Towards Robust Multimodal Open-set Test-time Adaptation via Adaptive Entropy-aware Optimization (ICLR 2025 · similarity 0.87)
- A Layer Selection Approach to Test Time Adaptation (AAAI 2025 · 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.
