STOCHASTIC DYNAMICS AND REPLICA SYMMETRY BREAKING OF NEUROSYSTEMS(Session I : Cross-Disciplinary Physics, The 1st Tohwa University International Meeting on Statistical Physics Theories, Experiments and Computer Simulations)
Yoshitake Yamazaki, Sujin Kim, Moyuru Ochiai · Institutional Repositories DataBase (IRDB) · 1996
Subject on neural networks (NNW) is very important, which is strongly concerned with nonequilibrium (NEQ) and/or equilibrium (EQ) statistical physics in random media, e.g. crystal growth, glasses, polymers and vortices. To establish foundations of such physics, stochastic dynamics and replica symmetry breaking (RSB) of generalized neurosystems are discussed using supersymmetry (SUSY) fields. The following items are concluded; (1) the valid range of replica symmetric (RS) solution, in the RSB-regime, (2) the validity of the hierarchical structure (HS), (3) the NEQ-behavior (e.g. aging) to a metastable state and (4) the partial annealing (PA) properties.