Research map / Bandits and online learning / Multi-armed bandits
Adaptive Multiple-Arm Identification
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 a hardness parameter that accounts for tolerated exchanges among near-boundary arms | AdaptiveTopK · ε-top-K arm identification | sample complexity: O(n/\sqrt{\epsilon}) vs. \Omega(n/\epsilon) | Helped |
| Remove the log n factor from the AdaptiveTopK sample-complexity bound | ImprovedTopK · ε-top-K arm identification | sample complexity: O((\log \epsilon^{-1} + \log \delta^{-1})H^{(t, \epsilon)}) vs. O((\log\log(\epsilon^{-1}) + \log n + \log\delta^{-1})H^ | Helped |
| Use the instance-dependent hardness parameter H^{(t, ε)} rather than H^{(0, ε)} | ImprovedTopK · ε-top-K arm identification | sample complexity: O((\log \epsilon^{-1} + \log \delta^{-1})H^{(t, \epsilon)}) vs. O((\log \delta^{-1} + \log H^{(0, \epsilon)})H^{(0, \eps | Helped |
| Allow a small number of accept/reject mistakes instead of requiring every classification to be correct | ImprovedTopK · ε-top-K arm identification | sample complexity: O((\log \epsilon^{-1} + \log \delta^{-1})H^{(t, \epsilon)}) | Helped |
| Use H^{(t, ε)} to characterize instance difficulty | ImprovedTopK · Worst-case ε-top-K arm identification | sample complexity: O\left(\frac{n}{\epsilon^2} \left(\log \epsilon^{-1} + \log \delta^{-1} \right) \right) vs. \Theta\left(\frac{n}{\epsilo | Mixed |
| Use an instance-dependent algorithm parameterized by H^{(t, ε)} | Any instance-dependent algorithm · Some instances with n = 2K and ε = Ω(n^{-1}) | sample complexity lower bound: H^{(t, \epsilon)} = \Theta(n); \Omega(n \log \epsilon^{-1}) | Hurt |
Closest papers by meaning
- Balancing Performance and Costs in Best Arm Identification (NeurIPS 2025 · similarity 0.91)
- Optimal Multi-Fidelity Best-Arm Identification (NeurIPS 2024 · similarity 0.91)
- On Universally Optimal Algorithms for A/B Testing (ICML 2024 · similarity 0.91)
- Breaking the $\log(1/\Delta_2)$ Barrier: Better Batched Best Arm Identification with Adaptive Grids (ICLR 2025 · similarity 0.90)
- Near Optimal Best Arm Identification for Clustered Bandits (ICML 2025 · similarity 0.90)
- Best Arm Identification for Stochastic Rising Bandits (ICML 2024 · similarity 0.90)
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.