Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Randomized Gaussian Process Upper Confidence Bound with Tighter Bayesian Regret Bounds
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 GP-UCB's log-growing beta_t with randomized zeta_t from two-parameter exponential distribution (IRGP-UCB) | GP-UCB · Holder table, Cross in tray, Ackley benchmarks | simple regret | Helped |
| Use two-parameter exponential randomization instead of Gamma randomization (RGP-UCB) | RGP-UCB · Benchmark functions | simple regret | Helped |
| Generalize RGP-UCB regret analysis to wider class of randomized confidence distributions (no distribution change) | RGP-UCB · Theory (finite and continuous domains) | BCR convergence rate | No clear effect |
| Two-parameter exponential randomization removes the need to increase confidence parameter with iterations | IRGP-UCB · Synthetic functions (theoretical confidence parameter) | confidence parameter value over iterations | Helped |
| Two-parameter exponential randomization (finite domain BCR bound) | IRGP-UCB · Finite domain theory | BCR convergence rate in T: O(sqrt(T gamma_T)) vs O(sqrt(T gamma_T log T)) | Helped |
| Two-parameter exponential randomization on continuous domain | IRGP-UCB · Continuous domain theory | BCR convergence rate in T | No clear effect |
| Set zeta via two-parameter exponential with s=d/2, lambda=1/2 in heuristic setting, versus Gamma-based RGP-UCB and GP-UCB heuristic beta_t | IRGP-UCB · Holder table function | simple regret | Helped |
| Two-parameter exponential randomization vs baselines on real materials datasets | IRGP-UCB · Perovskite, P3HT/CNT, AgNP datasets | simple regret | Mixed |
| Gamma-distributed randomized confidence (RGP-UCB) vs deterministic GP-UCB on synthetic functions | RGP-UCB · Synthetic functions | simple regret | No clear effect |
Closest papers by meaning
- Improved Regret Bounds for Gaussian Process Upper Confidence Bound in Bayesian Optimization (NeurIPS 2025 · similarity 0.94)
- Gaussian Process Upper Confidence Bound Achieves Nearly-Optimal Regret in Noise-Free Gaussian Process Bandits (NeurIPS 2025 · similarity 0.92)
- Posterior Sampling-Based Bayesian Optimization with Tighter Bayesian Regret Bounds (ICML 2024 · similarity 0.91)
- Improved Regret Analysis For Parallel Gaussian Process Bandit Optimization (NeurIPS 2026 · similarity 0.90)
- Improved Regret Analysis in Gaussian Process Bandits: Optimality for Noiseless Reward, RKHS norm, and Non-Stationary Variance (ICML 2025 · similarity 0.88)
- Minimizing UCB: a Better Local Search Strategy in Local Bayesian Optimization (NeurIPS 2024 · 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.