Reproduce “Multi-Fidelity Multi-Armed Bandits Revisited” 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
Xuchuang Wang, Qingyun Wu, Wei Chen, John C.S. Lui
The authors analyze multi-fidelity bandits, where higher-cost observations estimate arm rewards more accurately. For best-arm identification, they propose confidence-based selection with fidelity choices, under stated prior-mean and separation assumptions. For budgeted regret, an elimination algorithm achieves an O(K^(1/3)Λ^(2/3)(log Λ)^(1/3)) upper bound against an Ω(K^(1/3)Λ^(2/3)) worst-case lower bound, matching up to a logarithmic factor under the specified reward model.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.