Reproduce “Quasipolynomial-Time Planning for Decodable POMDPs” 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

Oliver Dobson, Varun Jog, Chun Hei Michael Shiu

The authors present a planning algorithm for finite-horizon partially observable Markov decision processes. It uses a state decoder correct with probability at least rho for every state, and tracks possible recent hidden exact reveals rather than assuming the system forgets history. For finite spaces, it returns a deterministic policy within additive epsilon of optimal in quasipolynomial time for fixed positive rho.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.