Reproduce “Improved Regret Bounds for Tracking Experts with Memory” 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
James Robinson, Mark Herbster
PoDS-θ and Share-θ address sequential prediction when useful experts change and return. They retain weight for experts with past performance, using projection or weight sharing, and run in linear time per trial. Both have an improved theoretical regret bound over the prior best-known bound under the paper’s realizability condition; the advantage over Fixed-Share for m ≤ k requires mild assumptions on n.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.