Character classification method and system based on brain-like computing platform

Abstract

The invention relates to a character classification method and system based on a brain-like computing platform. The method comprises the following steps that: a pulse neural network classification model is constructed; the spiking neural network model comprises a coding layer, a maximum pooling layer and a classification layer; the maximum pooling layer comprises a plurality of pooling neurons and a maximum pooling output layer; every two pooling neurons are connected with each other, and the pooling neurons are in suppression connection with each other; the pulse of the maximum pooling output layer is emitted by a first pulse neuron; the maximum pooling output layer comprises a neuron; the neurons are connected with each of the pooling neurons; and character classification is performed on the MNIST handwritten digital data set by utilizing the pulse neural network classification model. According to the invention, the character classification precision can be improved.

Description

technical field [0001] The present invention relates to the field of character classification, in particular to a character classification method and system based on a brain-inspired computing platform. Background technique [0002] In the artificial neural network, maximum pooling can extract texture features, reduce the influence of useless information, and obtain relatively lower dimensions by performing aggregate statistical processing on different features to avoid overfitting. That is, reduce the amount of parameters, reduce computational costs, and prevent overfitting. The ideal maximum pooling can extract useful information and discard useless information. In convolutional neural networks, maximum pooling has become an indispensable module in the network. Usually, it is only necessary to find the maximum value in the pooling area, that is, to determine It is the output of max pooling. However, it is still a difficult point to realize the maximum pooling in the spik...

AI Technical Summary

Problems solved by technology

However, it is still a difficult point to realize the maximum pooling in the spiking neural network. Since the output of the network is in the form of pulses, it is difficult to calculate the maximum output. In the spiking neural network, the maximum neuron changes dynamically over time. The comparison between neurons is difficult, and the brain-like hardware computing platform is difficult to compare information between neurons

In the classification tasks implemented based on the brain-inspired computing platform, the existing methods cannot meet the incomparable characteristics of the brain-inspired hardware computing platform, so these maximum pooling methods cannot be applied to the brain-inspired hardware computing platform

In the actual classification task, the pooling layer is essential, and the average pooling is usually used instead of the maximum pooling. This replacement method will lead to the loss of classification task accuracy, and finally the classification accuracy will decrease.

Images