An image classification method, storage device and processing device

Abstract

The invention belongs to the technical field of deep neural network, and specifically relates to a compression method of a deep convolutional neural network based on a small amount of unlabeled data, aiming to solve the problem of compressing all layers of a deep convolutional neural network on the basis of an existing full-precision network model. Compression without retraining based on the original labeled training data, the problem of large performance loss, including: obtaining the original deep convolutional neural network; sparse the weight tensor of each layer in the original deep convolutional neural network Operation to obtain multiple weight tensors containing more 0 elements; based on the compressed weight tensors, normalize the batch data in the compressed deep convolutional neural network through a small amount of unlabeled data The statistics in the layer are updated to obtain a new deep convolutional neural network. Through the embodiment of the present invention, the compression of a large-scale deep convolutional neural network that only relies on a small amount of unlabeled data is realized, and the performance loss is reduced.

Description

technical field [0001] The invention belongs to the technical field of deep neural networks, and in particular relates to an image classification method based on a small amount of unlabeled deep convolutional neural network compression. Background technique [0002] In recent years, convolutional neural networks have made great progress in target detection and recognition tasks, and their detection accuracy has reached the commercial level. At the same time, the rapid development of mobile terminals and smart devices has enabled researchers to see an opportunity to combine convolutional neural networks with portable devices. However, object recognition based on convolutional neural networks relies on large memory consumption and strong computing performance. They often rely on high-performance GPU devices, which are difficult to work on such as smartphones and embedded devices. Running a convolutional neural network model will quickly use up limited memory resources, hard ...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the above problems in the prior art, that is, to solve the performance loss when compressing all layers of the deep convolutional neural network on the basis of the existing full-precision network model, and not performing retraining based on the original labeled training data The larger problem, one aspect of the present invention, provides a method for compressing a deep convolutional neural network based on a small amount of unlabeled data, comprising the following steps:

Images