Neural network model compression method and device, storage medium and chip

Abstract

The invention provides a neural network model compression method in the field of artificial intelligence. The method comprises the steps that a server acquires a first neural network model uploaded byuser equipment and training data of a first neural network; obtaining a PU classifier according to the training data of the first neural network and the unmarked data stored in the server; using a PUclassifier to select extended data from the unmarked data stored in the server, the extended data having attributes and distribution similar to those of the training data of the first neural networkmodel; and training a second neural network model by utilizing a knowledge distillation KD method according to the extended data, taking the first neural network model as a teacher network model, andtaking the second neural network model as a student network model. And a PU classifier is adopted to select data with attributes and distribution similar to those of the training data of the first neural network model from the unmarked data, so that the compression accuracy of the neural network model is improved, and the transmission of a large amount of positive sample data is avoided.

Description

technical field [0001] The present application relates to the field of artificial intelligence, in particular to a method and device for compressing neural network models. Background technique [0002] Artificial intelligence (AI) is a theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is the branch of computer science that attempts to understand the nature of intelligence and produce a new class of intelligent machines that respond in ways similar to human intelligence. Artificial intelligence is to study the design principles and implementation methods of various intelligent machines, so that the machines have the functions of perception, reasoning and decision-making [0003] Computer vision is an integral part of various i...

AI Technical Summary

Problems solved by technology

Neural network model compression technology needs to provide a large amount of training data to make the network converge to a better result. However, it is very time-consuming for users to upload large amounts of data to the cloud, which will cause a decline in user experience

Some neural network model compression techniques only use a small amount of training data for model compression, but the compressed neural network model is difficult to achieve satisfactory results

Images