Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Convolutional neural network compression method and device and image classification method and device

A technology of convolutional neural network and compression method, which is applied in the image classification method and device, and the field of convolutional neural network compression, which can solve the problem of convolution kernel distribution sensitivity and achieve the effect of saving computing resources and time resource consumption

Pending Publication Date: 2021-11-16
INST OF AUTOMATION CHINESE ACAD OF SCI
View PDF0 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The present invention provides a convolutional neural network compression method and device, and an image classification method and device, which are used to solve the defect that the convolution kernel distribution is relatively sensitive in the prior art

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Convolutional neural network compression method and device and image classification method and device
  • Convolutional neural network compression method and device and image classification method and device
  • Convolutional neural network compression method and device and image classification method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0052] In order to make the purpose, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention , but not all examples. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0053]Current convolution kernel pruning methods are usually pruning schemes based on norm, geometric center similarity or structure search. In the norm-based method, it is necessary to set different pruning ratios according to the distribution of each convolutional layer, so as to avoid the accuracy drop caused by the different sensitivities of different layers to pruning. Improper pruning r...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention provides a convolutional neural network compression method and device and an image classification method and device, and the method comprises the steps: carrying out the convolution kernel clustering of each convolution operation layer in a pre-trained convolutional neural network, and obtaining a convolution kernel class cluster corresponding to each convolution operation layer; performing convolution kernel pruning on each convolution kernel class cluster corresponding to each convolution operation layer; and retraining the convolutional neural network after convolution kernel pruning to obtain a compressed convolutional neural network. According to the method, efficient convolution kernel pruning is achieved, meanwhile, the mode is not limited by different pruning sensitivities caused by convolution kernel distribution any more, and therefore consumption of computing resources and time resources of convolution kernel pruning is greatly reduced.

Description

technical field [0001] The present invention relates to the technical field of deep learning, in particular to a convolutional neural network compression method and device, and an image classification method and device. Background technique [0002] In recent years, deep neural networks, especially deep convolutional neural networks, have made great breakthroughs in the fields of computer vision, natural language processing, and speech recognition. With the improvement of the performance of deep convolutional neural networks, the number of parameters and computational complexity of the model have also increased significantly. Among them, the improvement of the number of parameters requires that the device storing the model has a large hard disk storage space. The increase in computational complexity causes the device to consume more power. This makes deep convolutional networks mostly only deployable on high-performance computer clusters. [0003] At the same time, with t...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06K9/62G06N3/04G06N3/08
CPCG06N3/082G06N3/045G06F18/23213
Inventor 王培松程健
Owner INST OF AUTOMATION CHINESE ACAD OF SCI
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products