The neural path probabilistic delay model

摘要

The neural path probabilistic delay (NPPD) algorithm models the neural network in a discrete-time manner analogous to the computer network. The four stages of the basic neuron cell component, i.e. neuron cell body, axon branches, synapse region, and dendrite trees are represented by the mathematical structures of cascaded Markov chains. A state probability transition matrix for the neuron cell is generated by a queueing model of signal interarrival and service rate probability density function (PDF) data. To obtain the delay value for the soma, the PDF functions respectively describing the rate of action potential generation at each cell body and neuron signal arrival at synaptic dendrite terminals, are subtracted and quantized horizontally. This quantization of service minus interarrival PDFs is analogous to the human nervous system encoding of action potential pulses. The specific state transition probability matrices associated with the axon, synapse, and dendrite stages are computed from experimental link structure data. Both PDF and link structure data are measured from neurons located in the specific section of the nervous system considered. An example of the NPPD measure is derived for a neural path in a segment of the cerebellar folium in the cerebellar cortex of the human brain, because a path in this area is initiated by a pain sensation stimulus to the cerebral cortex. Therefore, the motor response path delay parameters for the pain stimulus are of interest in the implementation of the robotic system.