Research map / Architectures and efficiency / Pruning and efficient tuning
Efficient Lottery Ticket Finding: Less Data is More
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 |
|---|---|---|---|
| Use PrAC sets instead of the full training set for iterative magnitude pruning | Lottery ticket finding across the reported CIFAR-10, CIFAR-100, and Tiny-ImageNet models · CIFAR-10, CIFAR-100, Tiny-ImageNet | training iterations: 83.93%; 69.53%; 82.85%; 63.54%; 92.77%; 74.92%; 76.14%; 86.56% saved | Helped |
| Combine training-critical examples (CET) with pruning-critical examples (CEP) (ablation) | PrAC lottery tickets · CIFAR-10, ResNet-20 | test accuracy: ∼1% | Helped |
| Add early stopping based on mask distance (ablation) | PrAC lottery tickets · CIFAR-10, ResNet-20 | training resources saved: 40.63% -> 83.93% | Helped |
| Add early stopping based on mask distance (ablation) | PrAC lottery tickets · CIFAR-10, ResNet-20 | accuracy loss: ≤ 0.50% | Mixed |
| Raise the forgetting threshold from 0 to 2 (ablation) | PrAC lottery tickets · CIFAR-10, ResNet-20, sparsity 16.78% | test accuracy: 91.05 -> 90.43 | Hurt |
| Raise the forgetting threshold from 0 to 4 (ablation) | PrAC lottery tickets · CIFAR-10, ResNet-20, sparsity 16.78% | test accuracy: 91.05 -> 90.35 | Hurt |
| Raise the forgetting threshold from 0 to 6 (ablation) | PrAC lottery tickets · CIFAR-10, ResNet-20, sparsity 16.78% | test accuracy: 91.05 -> 89.32 | Hurt |
| Raise the forgetting threshold from 0 to 8 (ablation) | PrAC lottery tickets · CIFAR-10, ResNet-20, sparsity 16.78% | test accuracy: 91.05 -> 89.08 | Hurt |
| Raise the forgetting threshold from 0 to 10 (ablation) | PrAC lottery tickets · CIFAR-10, ResNet-20, sparsity 16.78% | test accuracy: 91.05 -> 88.79 | Hurt |
| Transfer a PrAC set from ResNet-20 to another architecture | PrAC lottery ticket finding · CIFAR-10 and CIFAR-100, ResNet-56 and VGG-16 | test accuracy | No clear effect |
| Select PrAC examples rather than random, Entropy, or SVP subsets | Lottery ticket finding · CIFAR-10, ResNet-20 and other tested dataset-architecture combinations | test accuracy | Helped |
| Use PrAC-guided tickets rather than SynFlow, SNIP, or GraSP | CIFAR-10, ResNet-20, 93.13% sparsity · CIFAR-10 | test accuracy: +1.51%; +4.22%; +4.36% | Helped |
Closest papers by meaning
- The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks (ICLR 2019 · similarity 0.93)
- Plant 'n' Seek: Can You Find the Winning Ticket? (ICLR 2022 · similarity 0.92)
- The Elastic Lottery Ticket Hypothesis (NeurIPS 2021 · similarity 0.92)
- Coarsening the Granularity: Towards Structurally Sparse Lottery Tickets (ICML 2022 · similarity 0.92)
- Find A Winning Sign: Sign Is All We Need to Win the Lottery (ICLR 2025 · similarity 0.91)
- Fast Track to Winning Tickets: Repowering One-Shot Pruning for Graph Neural Networks (AAAI 2025 · similarity 0.90)
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.