Research map / Image generation / Image restoration
Towards Architecture-Agnostic Untrained Networks Priors for Image Reconstruction with Frequency Regularization
Yilin Liu, Yunkui Pang, Jiang Li, Yong Chen, Pew-Thian Yap
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 |
|---|---|---|---|
| replace white-noise input with Gaussian-blurred noise input (ablation) | A2_256 untrained network · fastMRI brain 4x | PSNR: 29.08 -> 33.34 | Improved |
| replace noise input with Fourier features of narrower bandwidth (L=4 vs L=16) (ablation) | A2_256 · fastMRI brain 4x | PSNR: 30.57 -> 31.92 | Improved |
| add Kaiser-based bandwidth-controllable upsampler (ablation) | deeper architectures (A8_64) · fastMRI brain 4x | PSNR: 33.14 -> 33.85 (A8_64); 31.70 -> 33.09 (A8_256) | Improved |
| add Kaiser upsampler on top of blur + Lipschitz for shallow wide network (ablation) | A2_256 · fastMRI brain 4x | PSNR: 32.90 -> 32.50 | No clear effect |
| add Lipschitz regularization alone (ablation) | untrained network · fastMRI brain 4x | PSNR: 29.41 -> 29.73 (A2_64); 29.08 -> 30.92 (A2_256) | Mixed |
| add TV regularization (ablation) | untrained network · fastMRI brain 4x | PSNR: 29.41 -> 29.61 (A2_64) | No clear effect |
| apply proposed frequency regularization to compact A2_64 | A2_64 · fastMRI knee 4x | PSNR: 27.62 -> 32.07 | Improved |
| add self-validation early stopping to the regularized network (ablation) | A2_64 (Ours) · fastMRI knee / Stanford FSE | runtime / PSNR: runtime 6.4 -> 0.3 mins; PSNR 32.07 -> 31.97 | Improved |
| apply early stopping to unregularized A2_64 (ablation) | A2_64 · fastMRI knee | PSNR: 27.62 -> 29.59, still below 32.07 | Mixed |
| increase upsampler attenuation from none to bilinear/Lanczos-style (ablation) | Deep Decoder · fastMRI brain 4x | PSNR: 24.55 -> 32.68 | Improved |
Closest papers by meaning
- UGoDIT: Unsupervised Group Deep Image Prior Via Transferable Weights (NeurIPS 2025 · similarity 0.88)
- Untrained Neural Nets for Snapshot Compressive Imaging: Theory and Algorithms (NeurIPS 2024 · similarity 0.86)
- Image Reconstruction Via Autoencoding Sequential Deep Image Prior (NeurIPS 2024 · similarity 0.86)
- Training the Untrainable: Introducing Inductive Bias via Representational Alignment (NeurIPS 2025 · similarity 0.86)
- ModHiFi: Identifying High Fidelity predictive components for Model Modification (NeurIPS 2025 · similarity 0.86)
- Breaking Information Isolation: Accelerating MRI via Inter-sequence Mapping and Progressive Masking (AAAI 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.
