Reproduce “Sharper Regret Bounds for Time-Varying Gaussian Process Bandits with Constant Exploration” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Matthias Mandl, Hanne Kekkonen
The authors analyze Gaussian-process upper-confidence-bound optimization for changing reward functions with costly, noisy evaluations. They discount past observations as the objective drifts and use a constant exploration weight with local confidence events. For a squared-exponential kernel, persistent-drift bounds give information gain per round of \(\widetilde{\mathcal O}(\epsilon^{1/2})\) and expected average regret of \(\widetilde{\mathcal O}(\epsilon^{1/4})\). These are long-horizon results, not fixed-horizon limits as drift vanishes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.