Research map / Bandits and online learning / Contextual bandits and Thompson sampling
A Robust Phased Elimination Algorithm for Corruption-Tolerant Gaussian Process Bandits
Ilija Bogunovic, Zihan Li, Andreas Krause, Jonathan Scarlett
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 |
|---|---|---|---|
| Use RGP-PE instead of the existing bound | GP bandits with the SE kernel · SE kernel | cumulative regret bound: \sqrt{T}(\log T)^{d}+C \sqrt{T} (\log T)^{d/2} -> \sqrt{T(\log T)^{d}}+C(\log T)^{3d/2} | Improved |
| Use RGP-PE instead of the existing bound when the existing scaling is non-trivial | GP bandits with the Matérn kernel · Matérn kernel | cumulative regret bound: T^{\frac{2\nu+3d}{4\nu+2d}}+CT^{\frac{\nu+d}{2\nu+d}} -> T^{\frac{\nu+d}{2\nu+d}}+CT^{\frac{3d}{4\nu+2d}} | Improved |
| Use RGP-PE instead of the existing bound | GP bandits with the linear kernel · Linear kernel | cumulative regret bound: \sqrt{Td}+Cd^{3/2} -> \sqrt{Td}+Cd^{3/2} | No clear effect |
| Use RGP-PE instead of GP-UCB under adversarial attacks | GP bandits · f1, Clipping, AggSub, Top-3, and Top-5 attacks | cumulative regret | Improved |
| Use RGP-PE instead of GP-UCB under adversarial attacks | GP bandits · Robot3D, Top-3 and Top-5 attacks | cumulative regret | Improved |
| Use RGP-UCB instead of RGP-PE | GP bandits under the tested attacks · f1 and Robot3D | cumulative regret | Mixed |
| Start the attack later rather than at the beginning | RGP-PE and RGP-UCB · f1 and Robot3D | cumulative regret | No clear effect |
Closest papers by meaning
- Robust Satisficing Gaussian Process Bandits Under Adversarial Attacks (NeurIPS 2025 · similarity 0.91)
- Corruption-Robust Linear Bandits: Minimax Optimality and Gap-Dependent Misspecification (NeurIPS 2024 · 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)
- Gaussian Process Upper Confidence Bound Achieves Nearly-Optimal Regret in Noise-Free Gaussian Process Bandits (NeurIPS 2025 · similarity 0.88)
- A Near-optimal, Scalable and Parallelizable Framework for Stochastic Bandits Robust to Adversarial Corruptions and Beyond (NeurIPS 2025 · similarity 0.88)
- Stochastic Bandits Robust to Adversarial Attacks (ICLR 2025 · similarity 0.87)
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.
