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

Convolutional neural network system and convolutional neural network quantification method

A convolutional neural network and convolutional technology, applied in the field of convolutional neural networks, can solve the problems of reduced accuracy of convolutional neural networks, affecting user experience, etc., to facilitate hardware design, reduce the amount of stored data, and improve accuracy and efficiency. Effect

Pending Publication Date: 2019-12-20
HUAWEI TECH CO LTD
View PDF4 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the current quantization method will cause the accuracy of the convolutional neural network to decrease and affect the user experience.

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 system and convolutional neural network quantification method
  • Convolutional neural network system and convolutional neural network quantification method
  • Convolutional neural network system and convolutional neural network quantification method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0034] The technical solution in this application will be described below with reference to the accompanying drawings.

[0035] First, some terms involved in this application are explained.

[0036] Quantization: Quantization is the process of mapping a set of numbers within the original value range to another target range through a mathematical transformation. Available methods such as table lookup, shift, truncation, etc. Among them, linear transformation is often used, and multiplication is usually used to complete this transformation.

[0037] Dequantization: The process of inversely transforming quantized numbers to the original range based on the previous linear transformation (quantization process). Dequantization can ensure that the system uses the quantized data to calculate according to a certain calculation rule. After dequantization, the result can still be very similar to the calculation result calculated by using the data within the original value range accordi...

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 system and a convolutional neural network quantization method, and the system comprises: a quantization module which is used for carrying out thequantization of the input data of an ith convolution layer of the system and the weight and bias of the ith convolution layer, and wherein i is a positive integer; and a convolution module that is used for carrying out convolution calculation on the quantized input data of the ith convolution layer, the quantized weight and the quantized bias to obtain a convolution result of the ith convolution layer. According to the convolutional neural network system provided by the invention, the weight and bias of the convolutional layer needing to be quantized in the convolutional layer and the input data input into the convolutional layer are quantized, and convolution calculation is performed by using the quantized input data, the quantized weight and the quantized bias so as to obtain the calculation result of each convolutional layer. The calculation amount of the convolutional neural network is reduced, and the quantization precision of the convolutional neural network is improved.

Description

technical field [0001] The present application relates to the field of convolutional neural networks, and more specifically, to a convolutional neural network system and a method for quantifying convolutional neural networks. Background technique [0002] The deep convolutional neural network has hundreds or even tens of millions of parameters after training, for example, the weight parameters and bias parameters included in the convolutional neural network model parameters, as well as the feature map parameters of each convolutional layer, etc. . And the storage of model parameters and feature map parameters is based on 32 bits. Due to the large number of parameters and the large amount of data, the entire convolution calculation process needs to consume a large amount of storage and computing resources. However, the development of deep convolutional neural networks is developing in the direction of "deeper, larger, and more complex". As far as the model size of deep conv...

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): G06N3/04
CPCG06N3/045G06N3/04
Inventor 郭鑫罗龙强余国生
Owner HUAWEI 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

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com