Modelling of neurone networks in Java and VHDL

This paper describes a model of neurones and networks of neurones based on their biological characteristics. Neurones communicate with each other at contact points called synapses that receive signals from axons of other neurones. Such a signal can be either excitatory or inhibitory. A neurone generally receives several input signals and produces an output signal depending on the weighted sum of the input signals. When the post-synaptic potential sum reaches the threshold for excitation, an action potential is generated and propagated along the axon. Previous work has postulated that artificial neurones that are more biological in their function can form more robust and noise-tolerant neurone networks compared with conventional artificial neural networks (ANN). We are therefore examining a proof-of-concept by modelling a simple neurone network that uses a waveform resembling a biological action potential. Our neurone network implementation is different from an ANN, in that each neurone is modelled on biological function and communicates via action potential pulses rather than static voltage summing networks. Note that we refer to neurone as having a biological style of behaviour as described in this paper and that we use the word neuron to designate the common neurons used in a typical ANN.