Data processing method, device and system based on neural network and storage medium

Abstract

The embodiment of the invention provides a data processing method, device and system based on a neural network and a storage medium. The method includes: acquiring to-be-processed data; and using theneural network to process the data to obtain a final processing result, wherein in a process of using the neural network to process the data, elements which are in a result matrix obtained by at leastone matrix multiplication operation and have absolute values which are of difference values from a set value and less than or equal to a threshold are set to the set value. The above technical solution can significantly simplify computing logic of the neural network under the premise of ensuring data processing accuracy, and correspondingly reduce logic gates in a multiplier of the neural network. Therefore, requirements on platform computing power and storage space aspects can be reduced, and then requirements on performance aspects of system power consumption, heat dissipation and the likecan be reduced.

Description

technical field [0001] The present invention relates to the field of artificial intelligence, and more specifically relates to a neural network-based data processing method, device, system and storage medium. Background technique [0002] Artificial neural network is a mathematical model based on imitating the structure and function of biological neural network, which can simulate and approximate functions for a certain type or some problems. Due to the development of big data, neural networks are currently widely used in the direction of massive databases such as image, speech, and natural language processing. [0003] At present, due to the accuracy requirements of the neural network, the complexity of the network is continuously increased and the spatial structure of the network is expanded. As a result, the amount of computation and the size of the network model are constantly increasing. On embedded platforms such as development boards and mobile phones, the limitatio...

AI Technical Summary

Problems solved by technology

As a result, the amount of calculation and the size of the network model are constantly increasing.

On embedded platforms such as development boards and mobile phones, limitations in computing power and storage space will pose great challenges to issues such as power consumption, area, and heat dissipation.

Images