It is well known that a sparsely coded network in which the activity level isextremely low has intriguing equilibrium properties. In the present work, westudy the dynamical properties of a neural network designed to store sparselycoded sequential patterns rather than static ones. Applying the theory ofstatistical neurodynamics, we derive the dynamical equations governing theretrieval process which are described by some macroscopic order parameters suchas the overlap. It is found that our theory provides good predictions for thestorage capacity and the basin of attraction obtained through numericalsimulations. The results indicate that the nature of the basin of attractiondepends on the methods of activity control employed. Furthermore, it is foundthat robustness against random synaptic dilution slightly deteriorates with thedegree of sparseness.
展开▼
