Part of Advances in Neural Information Processing Systems 14 (NIPS 2001)
Toshiyuki Tanaka, Shiro Ikeda, Shun-ichi Amari
We report a result of perturbation analysis on decoding error of the belief propagation decoder for Gallager codes. The analysis is based on infor- mation geometry, and it shows that the principal term of decoding error at equilibrium comes from the m-embedding curvature of the log-linear submanifold spanned by the estimated pseudoposteriors, one for the full marginal, and K for partial posteriors, each of which takes a single check into account, where K is the number of checks in the Gallager code. It is then shown that the principal error term vanishes when the parity-check matrix of the code is so sparse that there are no two columns with overlap greater than 1.