Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Adaptively Grouped Contextual Bandits for Heterogeneous Human-AI Decision Making with Conformal Prediction Sets
Yanchen Wu, Bo Li
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 adaptive grouping with counterfactual updates to Sequential Halving | SH · Synthetic (100 arms, 5-dim), binary | simple regret: 0.381 -> 0.298 | Improved |
| add adaptive grouping with counterfactual updates to Sequential Halving | SH · Synthetic (100 arms, 5-dim), continuous | simple regret: 0.211 -> 0.111 | Improved |
| add adaptive grouping with counterfactual updates to Sequential Halving | SH · Synthetic (100 arms, 10-dim), binary | simple regret: 0.421 -> 0.307 | Improved |
| add adaptive grouping with counterfactual updates to Sequential Halving | SH · Synthetic (100 arms, 10-dim), continuous | simple regret: 0.251 -> 0.113 | Improved |
| add adaptive grouping with counterfactual updates to Sequential Halving | SH · ImageNet16H-PS, continuous | simple regret: 0.205 -> 0.101 | Improved |
| use AGCB-SH instead of BOED in the small-arm regime | AGCB-SH · Synthetic (10 arms, 5-dim), binary | simple regret: 0.193 vs 0.190 | No clear effect |
| raise context dimension from 5 to 10 | AGCB-SH · Synthetic (100 arms), binary | simple regret: 0.298 -> 0.307 | No clear effect |
| add AGCB grouping on top of UCB with counterfactual updates | UCB (with CF) · ImageNet16H-PS, binary | cumulative regret: 93.97 -> 75.93 | Improved |
| add AGCB grouping on top of UCB with counterfactual updates | UCB (with CF) · ImageNet16H-PS, continuous | cumulative regret: 109.86 -> 107.24 | Improved |
| add AGCB grouping on top of SE with counterfactual updates | SE (with CF) · ImageNet16H-PS, binary | cumulative regret: 122.16 -> 93.87 | Improved |
| add AGCB grouping on top of SE with counterfactual updates | SE (with CF) · ImageNet16H-PS, continuous | cumulative regret: 143.87 -> 106.23 | Improved |
| add counterfactual updates to UCB | UCB · ImageNet16H-PS, binary | cumulative regret: 150.23 -> 93.97 | Improved |
Closest papers by meaning
- Diffusion Models Meet Contextual Bandits (NeurIPS 2025 · similarity 0.88)
- Learning Personalized Decision Support Policies (AAAI 2025 · similarity 0.87)
- MultiScale Contextual Bandits for Long Term Objectives (NeurIPS 2025 · similarity 0.87)
- Learning to Decide with AI Assistance under Human-Alignment (NeurIPS 2026 · similarity 0.87)
- Demystifying Online Clustering of Bandits: Enhanced Exploration Under Stochastic and Smoothed Adversarial Contexts (ICLR 2025 · similarity 0.86)
- Neural Dueling Bandits: Preference-Based Optimization with Human Feedback (ICLR 2025 · similarity 0.86)
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.
