Reproduce “Exploiting Higher Order Smoothness in Derivative-free Optimization and Continuous Bandits” 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
Arya Akhavan, Massimiliano Pontil, Alexandre B. Tsybakov
The authors analyze zero-order optimization with noisy function-value queries rather than derivatives. The method probes in opposite directions and weights the difference to cancel higher-order changes under smoothness. For strongly convex objectives, it establishes faster error rates with more iterations. At β=2, upper and lower bounds match in dimension and iteration count for fixed α, under distinct noise assumptions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.