Reproduce “Parameter Learning in Active Inference via Variational Message Passing” 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
Benedikt Waldeck
The method uses sum-product message passing to plan while learning unknown observation or transition matrices. It copies and freezes parameter priors for each imagined factor, preserving the link between an outcome and its one-step parameter update so information gain can shape choices. In grid tasks, SP-EFE records 300 goals and detour episodes, versus 146 ± 7 goals and no detours for SP-Pref-Plan, showing a task-specific result.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.