Research map / Bandits and online learning / Multi-armed bandits
A Biased Graph Neural Network Sampler with Near-Optimal Regret
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 Thanos instead of BanditSampler | GCN · Chameleon | test accuracy: 0.578 -> 0.607 | Helped |
| Use Thanos instead of BanditSampler | GCN · Squirrel | test accuracy: 0.383 -> 0.401 | Helped |
| Use Thanos instead of BanditSampler | GCN · ogbn-arxiv | test accuracy: 0.652 -> 0.663 | Helped |
| Use Thanos instead of BanditSampler | GCN · CoraFull | test accuracy: 0.555 -> 0.574 | Helped |
| Use Thanos instead of BanditSampler | GCN · ogbn-products | test accuracy: 0.754 -> 0.759 | Helped |
| Use Thanos instead of BanditSampler | GAT · Chameleon | test accuracy: 0.602 -> 0.620 | Helped |
| Use Thanos instead of BanditSampler | GAT · Squirrel | test accuracy: 0.386 -> 0.412 | Helped |
| Use Thanos instead of BanditSampler | GAT · ogbn-arxiv | test accuracy: 0.675 -> 0.680 | Helped |
| Use Thanos instead of BanditSampler | GAT · CoraFull | test accuracy: 0.544 -> 0.559 | Helped |
| Use Thanos instead of BanditSampler | GAT · ogbn-products | test accuracy: 0.756 -> 0.759 | Helped |
| Increase sample size k (3, 5, 10, 15) (ablation) | Thanos versus BanditSampler · ogbn-arxiv | test accuracy | Helped |
| Use Thanos instead of BanditSampler | GCN and GAT sampling · Cora | sampling approximation error | Helped |
Closest papers by meaning
- Bandit Samplers for Training Graph Neural Networks (NeurIPS 2020 · similarity 0.93)
- Graph Neural Network Bandits (NeurIPS 2022 · similarity 0.89)
- Shift-Robust GNNs: Overcoming the Limitations of Localized Graph Training data (NeurIPS 2021 · similarity 0.88)
- Efficient Learning of Linear Graph Neural Networks via Node Subsampling (NeurIPS 2023 · similarity 0.88)
- MuseGNN: Forming Scalable, Convergent GNN Layers that Minimize a Sampling-Based Energy (ICLR 2025 · similarity 0.87)
- How Graph Neural Networks Learn: Lessons from Training Dynamics (ICML 2024 · 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.