Pulse neural network conversion method and device based on pulse classification layer event driving

The invention discloses a pulse neural network conversion method and device based on pulse classification layer event driving. Neurons of a classification layer of a pulse neural network model converted by an artificial neural network model are quickly classified by adopting a pulse distribution priority judgment method; an artificial neural network model ReLU activation layer is replaced with neurons, so that an artificial neural network model which is more beneficial to the precision of a converted pulse neural network model is obtained; through weight normalization or threshold balance operation, conversion from the artificial neural network model to the pulse neural network model is completed; forward reasoning of the model is completed through information transmission of pulse signals transmitted by neurons in the pulse neural network; classification is carried out by using a pulse neural network classification layer event-driven judgment method, so that the reasoning speed can be improved to a certain e