Influence Of Synaptic Delay On Information Transmission Of Feedforward Neural Network

The processing and transmission of neural information in the nervous system is accomplished by a population of neurons that require a specific function.Artificial neural network is to study the related problems of information transmission and processing in the nervous system by constructing network models and computer simulation methods.The feedforward neural network is one of the simplest and most common models in artificial neural networks,and it is also an effective model for studying signal transmission between different regions of the brain.In this thesis,a 10-layer feedforward neural network is studied,and the effects of network topology,synaptic delay and neuron model parameters on system information transmission capability and processing mechanism are studied.The main work and conclusions are summarized as follows:1.The discharge characteristics and dynam:ic behavior of commonly used neuron models are studied.The response characteristics of neuron models to different inputs are simulated,and the advantages and disadvantages of various neuron models are analyzed.In addition,as a preparatory work,the effect of synaptic connection probabilities on the synchronization level of a single-layer neural network composed of excitatory and inhibitory neurons in a 4:1 ratio was studied.The results show that a large connection probability promotes nerves.Synchronization of the meta network.2.The problem of firing rate mode in a feedforward neural network composed of Hodgkin Huxley neurons and connected with a certain probability is systematically studied.The role of synaptic delay in this transmission mode is emphasized.The results show that the input noise intensity,interlayer connection probability,synaptic time constant and background noise intensity all have an impact on the firing rate transmission.Appropriate adjustment of these parameters can make the network have the ability to stably encode and transmit signals.Synaptic delay can both disrupt the normal transmission network and restore the normal transmission of the network that cannot be successfully transmitted.This is related to the strength of the time lag and the topology of the network.3.The problem of synfire mode transmission in the feedforward neural network composed of Leaky Integrate and Fire neurons and connected by global coupling is studied.The effects of synaptic connection strength and synaptic delay on transmission are mainly investigated.Under certain conditions,synaptic delay can help improve network transmission capacity.