Research map / Graph learning / GNN architectures
A Unified Lottery Ticket Hypothesis for Graph Neural Networks
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 |
|---|---|---|---|
| Jointly prune graph edges and model weights using UGS | GNNs across node classification and link prediction tasks · Small and large graph datasets | inference MACs at matched predictive performance: 20%~98% MACs saving on small graphs; 25%~85% MACs saving on large ones | Helped |
| Use UGS instead of random pruning | GNNs · Cora, Citeseer, and PubMed | test performance | Helped |
| Use UGS instead of ADMM at heavy graph sparsity | GNNs · Small and medium graph datasets | performance: 3~4% | Helped |
| Use GraphCL pre-trained initialization rather than random initialization | GCN GLTs · Node classification and link prediction on Cora and Citeseer | task performance | Helped |
| Randomly re-initialize the sparse model and learnable masks instead of rewinding to the original initialization (ablation) | GCN GLTs · Cora | accuracy: 79.70 -> 80.80; 78.50 -> 80.30; 75.70 -> 79.30; 63.70 -> 75.30 | Hurt |
| Increase graph pruning per UGS round from p_g=5% to p_g=20% (ablation) | UGS · GCN on Cora | accuracy | Hurt |
| Use p_g=10% instead of p_g=5% per UGS round (ablation) | UGS · GCN on Cora at higher sparsity levels | performance | No clear effect |
| Vary weight pruning per UGS round among p_theta=10%, 20%, and 40%, with p_g=20% (ablation) | UGS · GCN on Cora | performance | No clear effect |
| Prune the Ogbn-Proteins edge encoder weights along with other weights (ablation) | 28-layer ResGCN · Ogbn-Proteins node classification | performance | Hurt |
Closest papers by meaning
- Searching Lottery Tickets in Graph Neural Networks: A Dual Perspective (ICLR 2023 · similarity 0.96)
- Rethinking Graph Lottery Tickets: Graph Sparsity Matters (ICLR 2023 · similarity 0.96)
- Fast Track to Winning Tickets: Repowering One-Shot Pruning for Graph Neural Networks (AAAI 2025 · similarity 0.93)
- TEDDY: Trimming Edges with Degree-based Discrimination Strategy (ICLR 2024 · similarity 0.92)
- Weisfeiler and Leman Go Gambling: Why Expressive Lottery Tickets Win (ICML 2025 · similarity 0.90)
- Find A Winning Sign: Sign Is All We Need to Win the Lottery (ICLR 2025 · similarity 0.87)
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.