Research map / Bandits and online learning / Multi-armed bandits
Improving the Knowledge Gradient Algorithm
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 |
|---|---|---|---|
| Replace KG's reward based on the selected arm's mean with a best-arm-correctness reward | KG for best arm identification · Examples 1–3, Dose-finding, Drug Selection, Caption 853 | probability of false selection: Example 1: 0.29, 0.14 -> 0.21, 0.03; Example 2: 0.32, 0.13 -> 0.23, 0.03; Example 3: 0.14, 0.03 -> 0.09, 0.01 | Helped |
| Use one-step improvement in estimated correct-set probability for ε-good arm identification | ε-good arm identification · Examples 1–3, Dose-finding, Drug Selection, Caption 853 | probability of false selection: Example 1: 0.29, 0.07 -> 0.17, 0.03; Example 2: 0.35, 0.11 -> 0.29, 0.00; Caption 853: 0.21, 0.10 -> 0.10, 0.02 | Helped |
| Use one-step improvement in estimated correct-set probability for feasible-arm identification | Feasible arm identification · Examples 1–3, Dose-finding, Drug Selection, Caption 853 | probability of false selection: Example 1: 0.27, 0.16 -> 0.23, 0.02; Example 2: 0.33, 0.26 -> 0.24, 0.01; Dose-finding: 0.20, 0.17 -> 0.14, 0.01 | Helped |
| Set TTEI's probability of sampling the best arm to the optimal value β* | TTEI · Best arm identification | posterior convergence rate | Helped |
| Use equal allocation rather than a policy targeted at identifying the desired arms | Best, ε-good, and feasible arm identification · Synthetic and real examples | probability of false selection | Hurt |
Closest papers by meaning
- Optimal Batched Best Arm Identification (NeurIPS 2024 · similarity 0.86)
- Optimal Multi-Fidelity Best-Arm Identification (NeurIPS 2024 · similarity 0.86)
- Balancing Performance and Costs in Best Arm Identification (NeurIPS 2025 · similarity 0.86)
- Fixed Confidence Best Arm Identification in the Bayesian Setting (NeurIPS 2024 · similarity 0.84)
- Nearly Optimal Fixed-Confidence Best-Arm Identification with 1-Bit Feedback (NeurIPS 2026 · similarity 0.84)
- Almost Minimax Optimal Best Arm Identification in Piecewise Stationary Linear Bandits (NeurIPS 2024 · similarity 0.84)
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.