Research map / Optimization / Optimizers
GCN meets GPU: Decoupling “When to Sample” from “How to Sample”
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 LazyGCN recycling to nodewise sampling | GraphSAGE · Reddit | training time (s): 968.60 -> 43.85 | Helped |
| add LazyGCN recycling to nodewise sampling | GraphSAGE · Reddit | test F1: 93.26% -> 93.37% | No clear effect |
| add LazyGCN recycling to layerwise sampling | LADIES · Yelp | training time (s): 227.04 -> 77.62 | Helped |
| add LazyGCN recycling to subgraph sampling | GraphSAINT · Amazon | training time (s): 385.2 -> 198.4 | Helped |
| add LazyGCN recycling to nodewise sampling | GraphSAGE · Yelp | test F1: 62.78% -> 61.96% | Mixed |
| add LazyGCN recycling to small graphs | GraphSAGE/LADIES/GraphSAINT · Pubmed, Flickr | training time (s): Pubmed layerwise 0.50 -> 0.25 | Mixed |
| use very small recycling mini-batch size S (ablation) | LazyGCN · Reddit | validation F1 | Hurt |
| use very large recycle period R (ablation) | LazyGCN · Reddit | validation F1 convergence | Hurt |
| use growth rate rho>1 instead of rho<1 (ablation) | LazyGCN · Reddit | validation F1 convergence | Helped |
| fix inner-layer nodes during recycling instead of resampling them (ablation) | LazyGCN · Reddit | validation F1 and gradient MSE | Helped |
Closest papers by meaning
- IGLU: Efficient GCN Training via Lazy Updates (ICLR 2022 · similarity 0.91)
- FastGCN: Fast Learning with Graph Convolutional Networks via Importance Sampling (ICLR 2018 · similarity 0.90)
- Layer-Dependent Importance Sampling for Training Deep and Large Graph Convolutional Networks (NeurIPS 2019 · similarity 0.90)
- L2-GCN: Layer-Wise and Learned Efficient Training of Graph Convolutional Networks (CVPR 2020 · similarity 0.90)
- Generalization Guarantee of Training Graph Convolutional Networks with Graph Topology Sampling (ICML 2022 · similarity 0.90)
- MuseGNN: Forming Scalable, Convergent GNN Layers that Minimize a Sampling-Based Energy (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.