Nonresonant Parametric Amplification of Signals in Neural Networks Through the Rhythms of a Central Processor

Abstract

It is shown that giant nonresonant amplification of weak external signals is a possibility in neural networks. A mechanism underlying amplification is the modulation of the threshold and other neuron parameters. In biologic networks, cortical rhythms may serve as a modulation source. This work extends the Hopfield hypothesis that some cortical rhythms are related to signal amplification. In artificial neural networks, a special central generator may serve as a modulation source.