Research map / Label-efficient and robust learning / Test-time adaptation
Ground-Truth Free Meta-Learning for Deep Compressive Sampling
Xinran Qin, Yuhui Quan, Tongyao Pang, Hui Ji
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 |
|---|---|---|---|
| Use external meta-learning rather than training without meta-learning (ablation) | MetaCS · ADNI, radial masks, noiseless and noisy | PSNR (dB): 33.97 -> 34.85 | Improved |
| Add test-time model adaptation (ablation) | MetaCS · ADNI, radial masks, noiseless and noisy | PSNR (dB): 34.40 -> 34.85 | Improved |
| Use modified MAML rather than standard MAML (ablation) | MetaCS · ADNI, radial masks, noiseless and noisy | PSNR (dB): 34.01 -> 34.85 | Improved |
| Use iSURE instead of gSURE with Monte Carlo approximation (ablation) | MetaCS · ADNI, radial masks, noiseless and noisy | PSNR (dB): 32.57 -> 34.85 | Improved |
| Add null-space consistency loss during adaptation (ablation) | MetaCS · ADNI, radial masks, noiseless and noisy | PSNR (dB): 34.65 -> 34.85 | Improved |
| Adapt biases only rather than all model weights (ablation) | MetaCS · ADNI, radial masks, noiseless and noisy | PSNR (dB): 34.82 -> 34.85 | No clear effect |
| Adapt biases instead of feature-map gains (ablation) | MetaCS · ADNI, radial masks, noiseless and noisy | PSNR (dB): 34.58 -> 34.85 | Improved |
| Use ensemble inference with noise-perturbed inputs (ablation) | MetaCS · ADNI, radial masks | PSNR (dB) | No clear effect |
| Jointly use external learning and test-sample adaptation | MetaCS · MRI150, ADNI, and Set11 | PSNR (dB) | Improved |
| Train one model across noise levels rather than training individually | MetaCS · Set11, 25%, noise levels 1, 5, and 10 | PSNR (dB): 29.71 -> 29.64 | Hurt |
| Test on a measurement matrix different from the training matrix | MetaCS · ADNI, noiseless, Gaussian training mask 1/4 | PSNR drop: -0.07 | Hurt |
Closest papers by meaning
- MetaSCI: Scalable and Adaptive Reconstruction for Video Compressive Sensing (CVPR 2021 · similarity 0.88)
- Training Deep Learning Based Image Denoisers From Undersampled Measurements Without Ground Truth and Without Image Prior (CVPR 2019 · similarity 0.87)
- Reconstruction-free Cascaded Adaptive Compressive Sensing (CVPR 2024 · similarity 0.87)
- Multi-Cross Sampling and Frequency-Division Reconstruction for Image Compressed Sensing (AAAI 2024 · similarity 0.86)
- Untrained Neural Nets for Snapshot Compressive Imaging: Theory and Algorithms (NeurIPS 2024 · similarity 0.85)
- Using Powerful Prior Knowledge of Diffusion Model in Deep Unfolding Networks for Image Compressive Sensing (CVPR 2025 · 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.
