Data classification method and system based on convolutional neural network, medium and equipment

Abstract

The invention provides a data classification method and system based on a convolutional neural network, a medium and equipment. The data classification method comprises the steps: carrying out the preprocessing of obtained classification data, and constructing a data set; constructing a convolutional neural network, wherein the convolutional neural network at least comprises a convolutional layerused for extracting local features, and the convolutional layer compresses a feature matrix, and sparse matrix vector multiplication operation is performed on the generated sparse matrix on a graphicprocessing unit, and data in a data set is utilized to train the convolutional neural network; and preprocessing the to-be-classified data, inputting the preprocessed to-be-classified data into the trained convolutional neural network model, and outputting a data classification result. According to the data classification method, the feature matrix of the convolutional layer is compressed, and parallel computation is performed on the GPU, so that the memory consumption and zero value computation in the computation process are reduced, and the training time and memory consumption of the neuralnetwork are reduced.

Description

technical field [0001] The present disclosure relates to the technical field of data classification, and in particular to a convolutional neural network-based data classification method, system, medium and equipment. Background technique [0002] The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art. [0003] Convolutional neural networks are the cornerstone of deep learning in computer vision. At the same time, it has also achieved results in natural language processing, recommendation systems and speech recognition. In a convolutional neural network, the convolutional layer has the highest ratio of all layers. The larger the number of convolutional layers, the more information is obtained, the more features are extracted, and the model effect is better, but the calculation time and memory consumption ratio increase. Optimizing the convolutional layers in order to optimize the t...

AI Technical Summary

Problems solved by technology

[0005] The inventors of the present disclosure have found that in order to classify data, researchers have proposed a variety of convolution calculation schemes, such as im2col (Image to Column) algorithm, FFT algorithm and Winograd algorithm, etc., but the above convolution algorithms all use memory consumption and reduce The time scheme, although the calculation speed of convolution is improved, but the calculation memory is increased

Images