Data processing method and system and computer readable storage medium

Abstract

The embodiment of the invention discloses a data processing method and system and a computer readable storage medium. The method comprises the steps: processing to-be-analyzed data according to a dataprocessing mode through pre-built multi-path CNN network structures in the corresponding data processing mode; obtaining a processing result of the to-be-analyzed data according to the data processing modes and an operation calculation result of each CNN network structure, wherein the data processing modes are corresponding to the multi-path CNN network structures in a one-to-one manner, and allCNN network structures do not perform data interaction in a computing process. According to the invention, one FPGA is provided with the multi-path parallel CNN network structures, and different dataprocessing modes are corresponding to different network structures, thereby achieving the simultaneous parallel process of a large amount of data, saving the data processing time, and improving the work efficiency of a CNN algorithm. The method also can achieve the target recognition of the same target data through the multi-path CNN network structures, and facilitates the improvement of the target recognition accuracy of the CNN algorithm, thereby improving the overall performances of the CNN algorithm.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of deep learning, and in particular, to a data processing method, system, and computer-readable storage medium. Background technique [0002] With the rapid development of machine learning technology, deep learning, as a method based on representation learning of data in machine learning, has been widely used. [0003] At present, deep learning technology mainly takes DNN (Deep Neural Networks, deep neural network) as the research object. The mainstream of DNN for pattern recognition is the guided learning network, and the unguided learning network is more used for cluster analysis. For guided pattern recognition, since the category of any sample is known, the distribution of samples in space is no longer divided according to their natural distribution tendency, but according to the distribution of samples of the same type in space and the relationship between samples of different type...

AI Technical Summary

Problems solved by technology

[0005] The structure of CNN is relatively complicated. In order to improve the data processing speed of CNN, the existing technology mostly adopts CPU (Central Processing Unit, central processing unit) and FPGA (Field-Programmable Gate Array, Field-Programmable Gate Array) to form a heterogeneous acceleration platform for realization. Implement a CNN network structure on an accelerator board, but when there is a lot of data, or when the accuracy of target recognition is high, or when the data processing efficiency is high, the existing technology often cannot meet the actual requirements

Images