Reproduce “Gaussian Process Bandits for Constraint Black Box Lever Optimization” 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
Bram van den Akker, Hidde Fokkema, Anastasiia Kornilova
The authors apply GP-UCB to budget allocation, using kernels to share evidence across similar segment-weight configurations despite sparse feedback. In simulation, a Matérn + Linear model achieved regret of 30.9 ± 0.9 over 52 weekly decisions, versus 38.8 ± 1.9 for ε-greedy. A six-week live trial reported significant improvement but withheld absolute metrics; results do not establish performance against adversarial behavior.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.