Research map / Architectures and efficiency / Pruning and efficient tuning
Aligned Structured Sparsity Learning for Efficient Image Super-Resolution
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 |
|---|---|---|---|
| prune with ASSL (alignment + WN scale regularization) instead of training the small network from scratch (ablation) | EDSR baseline pruned at ratio 0.9 · Set5 x2 | PSNR (dB): 36.74 -> 37.23 | Helped |
| prune with ASSL instead of plain L1-norm pruning (ablation) | EDSR baseline pruned at ratio 0.9 · Set5 x2 | PSNR (dB): 36.87 -> 37.23 | Helped |
| prune with ASSL instead of plain L1-norm pruning (ablation) | EDSR baseline pruned at ratio 0.1 · Set5 x2 | PSNR (dB): 37.91 -> 37.94 | Helped |
| add sparsity structure alignment (ASSL) vs L1-norm pruning with misaligned residual channels, one epoch finetuning (ablation) | EDSR baseline pruned at ratio 0.1 · Set5 x2 | PSNR (dB): L1-norm drops 3.29 dB vs ASSL drops 0.15 dB | Helped |
| ASSL pruning vs scratch training at moderate pruning ratio (ablation) | EDSR baseline pruned at ratio 0.5 · Set5 x2 | PSNR (dB): 37.75 -> 37.82 | Helped |
| use ASSLN instead of IMDN (best prior lightweight SR) | lightweight SR · Urban100 x4 | PSNR (dB): 26.04 -> 26.27 | Helped |
| use ASSLN pruning instead of NAS/KD compression (FALSR-A, CARN+KD) | lightweight SR · Set5 x2 | PSNR (dB): 37.82 -> 38.12 | Helped |
Closest papers by meaning
- Learning Efficient Image Super-Resolution Networks via Structure-Regularized Pruning (ICLR 2022 · similarity 0.94)
- Lightweight Image Super-Resolution via Flexible Meta Pruning (ICML 2024 · similarity 0.93)
- Structured Sparsity Learning for Efficient Video Super-Resolution (CVPR 2023 · similarity 0.91)
- Image Processing GNN: Breaking Rigidity in Super-Resolution (CVPR 2024 · similarity 0.87)
- Learning Coupled Dictionaries from Unpaired Data for Image Super-Resolution (CVPR 2024 · similarity 0.86)
- Adaptive Sharpness-Aware Pruning for Robust Sparse Networks (ICLR 2024 · similarity 0.86)
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.