Research map / Architectures and efficiency / Pruning and efficient tuning
Memory-Friendly Scalable Super-Resolution via Rewinding Lottery Ticket Hypothesis
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 rewinding to iterative magnitude pruning (ablation) | EDSR · Set14, 4× | PSNR (dB): 27.66 -> 28.12 | Helped |
| Add stochastic self-distillation to IMP with rewinding (ablation) | EDSR · Set14, 4× | PSNR (dB): 28.12 -> 28.33 | Helped |
| Add SSD to the pruned model (ablation) | EDSR at 94% sparsity · Set14, 2×, 3×, and 4× | PSNR (dB): 32.88 -> 33.09; 29.73 -> 29.92; 28.12 -> 28.33 | Helped |
| Compress EDSR by pruning 94% of parameters with IMP-SSD | EDSR · Set14, 2×, 3×, and 4× | PSNR (dB): 33.63 -> 33.09; 30.30 -> 29.92; 28.55 -> 28.33 | Hurt |
| Compress EDSR by pruning 88% of parameters with IMP-SSD | EDSR · Set5, Set14, B100, and Urban100, 4× | PSNR (dB): 32.09 -> 31.89; 28.55 -> 28.44; 27.54 -> 27.46; 25.97 -> 25.71 | Hurt |
| Compress EDSR by pruning 94% of parameters with IMP-SSD | EDSR · Set5, Set14, B100, and Urban100, 4× | PSNR (dB): 32.09 -> 31.74; 28.55 -> 28.33; 27.54 -> 27.39; 25.97 -> 25.48 | Hurt |
| Use the backward-stage scalable model rather than separate IMP-SSD subnetworks (ablation) | EDSR · Set5, Set14, B100, and Urban100, 4× | PSNR (dB): 31.89 -> 31.82; 28.44 -> 28.40; 27.46 -> 27.43; 25.71 -> 25.61 | No clear effect |
| Increase EDSR sparsity to 99% (ablation) | EDSR IMP-SSD · Set14, 4× | PSNR (dB): 28.09 -> 20.36 | Hurt |
| Prune the EDSR model to 94% sparsity with IMP-SSD | EDSR · Set14, 4× | PSNR (dB): 28.55 -> 28.33 | Helped |
Closest papers by meaning
- SAVSR: Arbitrary-Scale Video Super-Resolution via a Learned Scale-Adaptive Network (AAAI 2024 · similarity 0.86)
- Lightweight Image Super-Resolution via Flexible Meta Pruning (ICML 2024 · similarity 0.86)
- Edge-SD-SR: Low Latency and Parameter Efficient On-device Super-Resolution with Stable Diffusion via Bidirectional Conditioning (CVPR 2025 · similarity 0.85)
- Image Processing GNN: Breaking Rigidity in Super-Resolution (CVPR 2024 · similarity 0.85)
- TSD-SR: One-Step Diffusion with Target Score Distillation for Real-World Image Super-Resolution (CVPR 2025 · similarity 0.85)
- Efficient Attention-Sharing Information Distillation Transformer for Lightweight Single Image Super-Resolution (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.