Vector Fields and Neural Networks
R. Vilela Mendes, J. Taborda Duarte · Complex Systems · 1992
We consider neural network models described by syste ms of (continuous t ime) differential equa tions. Th e dynamical na ture of each model is identified , symmet ric networks being relat ed t o gra dient vector fields and asymmet ric networks decomposed int o t heir gradi ent and Hamilt onian compo nents . From thi s identificati on follows, in particular , a simple charac terization of st ructural stability for sym metric networks and a limit cycle analysis of asymmetric networks as generators of coherent tempor al pattern s.