Reproduce “Learning to Decode Linear Codes with Generalized Belief Propagation Network” 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
Samira Malek
GBPNet extends learned belief propagation, retaining the Tanner graph and symmetry that permit training on one codeword. It adds cascaded variable-node updates and learned message scaling. In tests, it lowers bit error rates: on MacKay (96,48) at 5 dB, its −ln(BER) score is 13.11, versus 11.77 for CrossMPT and 9.40 for conventional BP, within tested codes and settings.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.