Research On Robustness Ofcomplex Spiking Neural Networkbased On Chemical Synapticregulation Mechanism

At present,the spiking neural network model has the problem of insufficient biological rationality.The delay of synaptic transmission plays a key role in the neural circuit,which can significantly improve the ability of the neural network to process information.The construction of spiking neural network with synaptic delay can improve the biological rationality of the network model.The complex spiking neural network which is based on the chemical synaptic regulation model with time delays is constructed.Based on the firing rate and correlation between membrane potential,the robustness of complex spiking neural networks under noise and different conditions of neuron injury are analyzed to explore their anti-interference and anti-injury ability,which has important theoretical value for improving the information processing ability of neural network.The main works of this paper are as follows:1.In order to explore the anti-interference ability of scale-free spiking neural network,the anti-interference ability of high cluster scale-free spiking neural networks with random time delays based on the firing rate and correlation between membrane potential under different intensities of white Gaussian noise,impulsive noise,and electric field noise,and it is compared with the anti-interference ability of high cluster scale-free spiking neural networks with fixed time delays and no time delays.The anti-interference ability of high cluster scale-free and low cluster BA scale-free spiking neural networks with random time delays are compared and analyzed under different noises.The correlation between the regulation of synaptic plasticity and anti-interference ability of the network is analyzed.The experimental results show that the high cluster scale-free spiking neural network with random time delays,fixed time delays and no time delays has an anti-interference effect on a certain intensity of noise,and the anti-interference ability of the high cluster scale-free network with random time delays is the best.The high cluster scale-free and low cluster BA scale-free spiking neural networks with random time delays have an anti-interference effect on a certain intensity of noise,and the anti-interference ability of high cluster scale-free networks is better than the low cluster BA scale-free network.The anti-interference ability of network is closely related to the regulation of synaptic plasticity.2.In order to explore the anti-injury ability of complex spiking neural network under random and deliberate attacks,based on the firing rate and correlation between membrane potential,the anti-injury ability of high cluster scale-free spiking neural network with random time delays are analyzed under different conditions of neuron injury,and the anti-injury ability of small world spiking neural network is compared.The correlation between the regulation of synaptic plasticity and anti-injury ability of the network is analyzed.The experimental results show that the high cluster scale-free spiking neural network networks have a certain anti-injury ability against random and deliberate attacks.Compared with deliberate attacks on the low and medium degree nodes and random attacks,the anti-injury ability of spiking neural network reduces significantly under deliberate attacks on high degree nodes.The anti-injury ability of high cluster scale-free spiking neural network is better than the small world spiking neural network under random attacks.The anti-injury ability of small world spiking neural network is better than the scale-free spiking neural network under deliberate attacks on the high degree nodes.The anti-injury ability of network is closely related to the regulation of synaptic plasticity.