Implementation method of end-to-end functional pulse model based on spiking neural network

Abstract

The invention discloses an implementation method of an end-to-end functional pulse model based on a spiking neural network, which relates to the technical field of artificial intelligence and comprises the following steps: defining an input pulse binary group Di and an expected output pulse binary group Dd; constructing a functional pulse model based on the functional pulse response function and the dynamic synaptic function, initializing parameters of the functional pulse model, and setting a training round number epochmax; constructing a loss function of the functional pulse model, and calculating a training loss value L according to the loss function; when the training loss value L is not equal to zero and the current training round number is smaller than epochmax, training the functional pulse model by using a back propagation algorithm, and updating parameters until the functional pulse model converges to complete training; the trained functional pulse model is tested, if the requirement is met, the trained functional pulse model is output, the time sequence learning task of the method is obviously superior to that of a traditional pulse neural network, the classification accuracy of the model is improved, and the training speed of the model is increased.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, and more specifically relates to a realization method of an end-to-end functional impulse model based on impulse neural network. Background technique [0002] Spiking Neural Network (SNN) belongs to the third-generation neural network model, which achieves a more advanced level of biological neural simulation. In addition to neuron and synaptic states, SNNs incorporate the concept of time into their operations. SNNs aim to bridge the gap between neuroscience and machine learning, using models that best fit biological neuronal mechanisms for computation. Spiking neural networks are fundamentally different from current popular neural network and machine learning methods. SNNs use spikes -- discrete events that occur at points in time -- rather than the usual continuous values. Each spike is represented by differential equations for biological processes, the most important...

AI Technical Summary

Problems solved by technology

[0017] The purpose of the present invention is: in order to solve the time encoding method in the existing spiking neural network model, each simulation data is converted into a specific time period, and when the input has high dimensions, the time series has its own time characteristics, most of the encoding In this case, the problem of not being effective enough, the present invention provides a method for realizing the end-to-end functional impulse model based on the impulse neural network

Images