A Computational Friendly Improved LIF Model Considering Refractory Period

In the past few years, spike neural networks (SNN) have been called the third generation neural networks. Compared with traditional neural networks, SNN has been widely concerned because of its space-time characteristics, diverse coding methods, and character of event driven. In order to make the spiking neural network show more complex behaviors, this paper improved the leakage integrity fire (LIF) model, considered the refractory period, proposed the refractory leakage integrity fire (RLIF) model, and simulated the neural network behavior. Through the regulation of various factors of SNN, the experiment is carried out to observe the different behaviors emerging from SNN, so as to analyze the influence of various factors in the model on SNN emerging behavior, and provide basis for discovering the intelligent behavior of SNN and applying it. The following research can further analyze the impact of neural network structure and training methods of spiking neural network based on RLIF model.