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

Convolutional neural network model based data processing method and device

A technology of a convolutional neural network and a data processing device, which is applied in the field of data processing based on a convolutional neural network model, and can solve problems such as decreased computing performance.

Active Publication Date: 2017-11-24
INSPUR SUZHOU INTELLIGENT TECH CO LTD
View PDF4 Cites 11 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In the prior art, the common method of model compression is to prune the model (that is, delete the smaller parameters in the model or the parameters that meet certain conditions), and store the parameters in the form of a sparse matrix, so that although the compression effect, but the accuracy loss of the model is also inevitable; in addition, there are also compression methods that adopt the method of retraining the cropped model to reduce the loss of model accuracy, but the computing performance when using the model to reason and predict is significantly reduced

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 model based data processing method and device
  • Convolutional neural network model based data processing method and device
  • Convolutional neural network model based data processing method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0020] Hereinafter, the present invention will be described in detail with reference to the drawings and examples. It should be noted that, in the case of no conflict, the embodiments in the present application and the features in the embodiments can be combined with each other.

[0021] The core principle of the present invention: the parameter data of the convolutional layer and the fully connected layer in the pre-trained convolutional neural network model are cyclically retrained as discrete data in the preset format, and the converted model has no loss of precision; the preset format Discrete data is stored in low bits.

[0022] Among them, the parameter data of the convolutional layer and the fully connected layer in the model are all in FP32 format (32-bit floating-point storage format).

[0023] Further, before the parameter data of the convolutional layer and / or the fully connected layer in the convolutional neural network model are cyclically trained to obtain the d...

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 model based data processing method and device. The method includes steps: cyclically training parameter data of a convolutional layer and / or a full connection layer in a convolutional neural network model to acquire discrete data in a preset format; storing the discrete data in the preset format by preset bits. According to the technical scheme, the parameter data are converted into the discrete data which are then stored by the preset bits, and consequently compression storage of the model is realized while accuracy loss of the model after conversion is avoided; due to adoption of the discrete data in the preset format, operation efficiency is greatly improved.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a data processing method and device based on a convolutional neural network model. Background technique [0002] Human beings are currently in the tide of the fourth industrial revolution, and artificial intelligence is the key technology leading this tide. Due to the powerful functions of artificial intelligence technology and broad application scenarios, it is expected to bring breakthroughs to all walks of life in the future and penetrate into all aspects of life. Therefore, scientists, researchers, companies, and online communities from all over the world are vigorously researching and promoting the development of artificial intelligence, and deep learning is one of the most popular technologies: deep learning generally uses neural network models, and uses a large amount of data to analyze the neural network. The model is trained so that the machine can learn ...

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
Patent Type & Authority Applications(China)
IPC IPC(8): H03M7/30G06N3/04G06N3/08
CPCH03M7/30G06N3/08G06N3/045
Inventor 谢启凯吴韶华
Owner INSPUR SUZHOU INTELLIGENT TECH CO LTD
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