Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Offline Neural Contextual Bandits: Pessimism, Optimization and Generalization
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 NeuraLCB with pessimistic lower-confidence-bound decisions instead of the compared baselines | Offline contextual bandits · Synthetic and real-world OPL tasks | sub-optimality | Helped |
| Use a trainable neural representation instead of linear models or fixed neural features | Offline contextual bandits · Synthetic datasets | sub-optimality | Helped |
| Add pessimistic lower-confidence-bound decisions to the trained neural model (ablation) | NeuraLCB versus NeuralGreedy · Real-world datasets | sub-optimality | Helped |
| Use S-mode single-example updates instead of B-mode batch updates (ablation) | NeuraLCB · Real-world datasets | sub-optimality | Helped |
| Replace NeuraLCB with LinLCB, NeuralLinLCB, NeuralLinGreedy, or KernLCB | Offline contextual bandits · Most synthetic and real-world tasks | sub-optimality | Hurt |
| Use LinLCB or KernLCB instead of NeuraLCB | Offline contextual bandits · Mushroom | sub-optimality | No clear effect |
| Use KernLCB instead of NeuraLCB | Offline contextual bandits · Adult | sub-optimality | Mixed |
Closest papers by meaning
- Off-Policy Learning in Large Action Spaces: Optimization Matters More Than Estimation (ICML 2026 · similarity 0.92)
- Direction-Aware Offline-to-Online Learning in Linear Contextual Bandits (NeurIPS 2026 · similarity 0.90)
- Selective Uncertainty Propagation in Offline RL (AAAI 2025 · similarity 0.90)
- Leveraging Offline Data in Linear Latent Contextual Bandits (ICML 2025 · similarity 0.90)
- Contextual Bandits with Online Neural Regression (ICLR 2024 · similarity 0.90)
- Leveraging (Biased) Information: Multi-armed Bandits with Offline Data (ICML 2024 · similarity 0.89)
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.