中文摘要：

目的 采用脉冲耦合神经网络（pulse coupled neural network，PCNN）对海马CA3区神经元集群放电进行仿真研究。方法 PCNN模型由120个神经元组成，兴奋性神经元与抑制性神经元个数之比为5：1。神经元间的连接权重为高斯分布。结果PCNN仿真模型输出结果表明，在周期信号、Gaussian随机信号及两类信号的线性叠加三类输入模式下，PCNN仿真网络的输出平均发放率均小于10％；神经元之间的稀疏连接可以通过调节权重实现。结论 ①在三类不同输入模式下，PCNN仿真网络的输出平均发放率均小于10％，满足海马CA3区神经元稀疏编码的特点。②在不同刺激下，模型中神经元的平均放电频率为6．02＋1．55Hz，其频率范围为3．6-8．6Hz，与海马区神经元放电的特征频率（θ节律）一致。③在PCNN仿真模型中，神经元之间的连接可通过调节权重矩阵实现，满足海马CA3区神经元稀疏连接的特点。④针对不同的输出模式，PCNN仿真网络可输出网络中每个神经元在不同时刻放电的时间序列。PCNN仿真模型可以反映海马CA3区神经元集群的放电特性，其仿真结果可以作为研究海马区神经元集群编码的仿真数据。

英文摘要：

Objective To simulate firings in hippocampus CA3 area with pulse coupled neural network （PCNN）. Methods The model consists of 120 neurons, of which the ratio of excitatory to inhibitory neurons is 5：1. The weight among neurons in the ensemble is set according to Gaussian distribution. Results Results show that for three different inputs of sinusoidal, phase random and the sum of the above two inputs, average population firings rate is less than 10%; the sparse connectivity among neurons can be adjusted by weight matrix. Conclusion ①Under three different types of inputs, the mean activity level of PCNN is less than 10%, which satisfies the sparse coding of hippocampus CA3. ②Mean firing rate of PCNN model is 6.02± 1.55Hz （range 3.6-8.6Hz）, which is consistent with hippocampus characteristic rhythm （0 rhythm）. ③The connectivity of the neurons is adjusted by the synaptic weight matrix. It satisfies the sparse connectivity of hippocampus neuron.④PCNN outputs give out different time series of firing according to the inputs and reflect the characteristic population firing of hippocampus CA3 area; which may be further used in future coding studies.