Reproduce “Nonlocal Transition Kernel for Efficient 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
Kaiji Sekimoto, Muneki Yasuda
The proposed kernel improves sampling for restricted Boltzmann machines by carrying states up a learned RBM sequence, updating at the top, and returning in one transition. Its joint target distribution is invariant. On GenRBM, learning appears stable at depth three; across four datasets, median total-variation distance falls faster than with BGS or deep tempering, without establishing faster mixing or lower computation costs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.