Reproduce “Strategic PAC Learnability via Geometric Definability” 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
Yuval Filmus, Shay Moran, Elizaveta Nesterova, Nir Rosenfeld, Alexander Shlimovich
The authors study strategic learnability when an input is positive if it can reach a point classified positive. They show that a real-line class with VC dimension 1 can induce infinite VC dimension under every fixed-radius interval rule. They also prove PAC learnability when classifiers and reachability are first-order definable in R_exp, though broad guarantees are qualitative.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.