Reproduce “Latent Kullback Leibler Control for Continuous-State Systems using Probabilistic Graphical Models” 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
Takamitsu Matsubara, Vicenç Gómez, Hilbert J. Kappen
The authors present a method for controlling systems by learning discrete latent states from exploratory trajectories, solving a KL-control problem in that representation, and mapping the result to commands. An HMM supports exact control, while a factorial HMM scales through approximate control. In simulated reaching, AVKL error stays approximately constant as chain count grows under factorized uncontrolled joint dynamics.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.