Reproduce “Conformal Sequential Decision Making With Cost and Sublinear Policy Regret Guarantees” 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
Jayadev Naram, Giuseppe Durisi, Matteo Zecchin, Osvaldo Simeone
OCEA learns candidate compression thresholds across blocks and switches to a zero-cost action when a block’s remaining distortion budget cannot cover another round. In a synthetic compression experiment, normalized policy regret for OCEA decreases with horizon, while OCSC’s remains bounded away from zero. The method guarantees cost compliance in every episode and expected sublinear policy regret against its feasible comparator.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.