Reproduce “Equilibrium and non-Equilibrium regimes in the learning of Restricted Boltzmann Machines” 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
Aurélien Decelle, Cyril Furtlehner, Beatriz Seoane
The authors study how MCMC chain length affects training and generation in restricted Boltzmann machines (RBMs). They relate the training length k to the model’s changing mixing time, which determines whether samples reflect equilibrium behavior. On MNIST, Rdm-k models show their best sample quality around generation time t_G ∼ k; this short-run result does not establish that the equilibrium distribution matches the data.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.