A neural network model compression method and device

Abstract

The present invention provides a method and device for compressing a neural network model, wherein the method includes: inputting training data into a neural network model to be compressed and a target neural network model; extracting feature vectors and classifying the training data based on the neural network model to be compressed As a result, the target neural network model is trained to obtain a compressed neural network model; wherein, the number of parameters of the target neural network model is less than the number of parameters of the neural network model to be compressed. The embodiment of the present invention guides the target neural network model to train based on the feature vectors and classification results extracted from the training data by the neural network model to be compressed, and the finally obtained compressed neural network model and the classification result of the same training data by the neural network model to be compressed are In the same way, there will be no loss of accuracy in the process of model compression, and the size of the model can be compressed under the premise of ensuring accuracy to meet the dual requirements for accuracy and model size.

Description

technical field [0001] The present invention relates to the technical field of machine learning, in particular to a neural network model compression method and device. Background technique [0002] With the rapid development of neural networks in the fields of image, voice, and text, it has promoted the landing of a series of intelligent products. In order to allow the neural network to better learn the characteristics of the training data to improve the model effect, the parameters used to represent the neural network model increase rapidly, and the number of layers of the neural network continues to increase, resulting in a large number of parameters in the deep neural network model. Model training and application Insufficient amount of calculation in the process; this leads to the fact that most of the products based on neural network rely on the drive of server-side computing power, and are very dependent on a good operating environment and network environment, which lim...

AI Technical Summary

Problems solved by technology

[0004] However, the above-mentioned methods directly perform model compression on the model to be compressed, and perform model compression on the premise of sacrificing the accuracy of the model, which often cannot meet the use requirements for accuracy.

Images