Research On Synchronization Of Neuronal Networks With Varied Weights

Neurons are basic structural and functional units of nervous systems,and are also basic nodes of the neural networks of human brain.In nervous systems,the processing of information is completed by clusters of neurons,in which the synchronization of these neurons is thought to play an important role.The strengths of synaptic connections among neurons are not constant,but are always changing with activities like growth,learning,memorizing and so on,hence biological neural systems with varied weights are multi-level information networked systems.Therefore,it is necessary to study the synchronization of neural networks with varied weights,which can help us to explore the information processing mechanisms of the brain.The according research has a positive theoretical and practical significance for the treatment of diseases of the nervous system and development of artificial intelligence.In this thesis,the dynamic properties of neural networks with varied weights are studied.We mainly analyze the influence of various systematic parameters on the synchronization of the neuronal networks under different learning rules.Firstly,the classical Hodgkin-Huxley(HH)neuronal model is improved.A calcium channel is introduced into the HH neuronal model,which makes the firing pattern of the neuronal model change from spiking to bursting.Secondly,the synchronization of the excitatory and inhibitory neuronal networks based on the improved HH neuronal model under the Oja rule were studied respectively.The effects of coupling strength,synaptic learning rate,time delay and decay time on the dynamic properties of the networks are studied by using the characteristic measurements such as the degree of excitation,the mean phase difference and the frequency of oscillation.In excitatory neuronal networks,thecoupling strength and the learning rate have a great influence on the dynamic properties.In a certain range,the increase of the coupling strength could decrease the degree of excitement and improve the synchronization of the networks.The increase of the learning rate could improve the degree of excitement at the beginning and then decrease it.But increasing the learning rate could decrease the degree of excitement at the beginning and then improve it.In inhibitory neuronal networks,the coupling strength and the learning rate have little effect on dynamic properties of the networks.In addition,the effects of the systematic parameters on the dynamic properties in the Izhikevich neuronal networks under spike-timing-dependent(STDP)learning rule are studied.The influence of the recovery parameter on the synchronization of the network is mainly studied.It is found that the recovery parameter has a very significant effect on the firing rhythm of the neurons.The results of the research are explained from biological views.It can be seen as a reference for revealing the information transmission mechanisms of neurons.