Pulse propagation networks (PPN) are neural networks in which individual action potentials encode information. The dynamics of PPN depend not only on the synaptic weights of connections but also the delay in the propagation of action potentials between neural elements. It is known that PPN can perform complex computations and information processing by encoding information as the time intervals between action potential events. In this paper we approach the practical question of constructing PPN to generate, recognize and learn arbitrary recurrent signals. We present specific examples of networks that generate and recognize signals and also describe a learning algorithm that allows PPN to learn by self-organization. Finally we discuss the possible importance of dynamical fluctuations about the mean-activity field of a neural network.
展开▼
