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

Depth learning algorithm achieving method and platform based on domestic many-core processor

A many-core processor and deep learning technology, applied in the field of deep learning algorithm implementation methods and platforms, can solve the problems of complex construction and long model training time, and achieve the effect of wide application prospects, prominent substantive characteristics, and reliable principles.

Inactive Publication Date: 2017-08-22
ZHENGZHOU YUNHAI INFORMATION TECH CO LTD
View PDF2 Cites 13 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to solve the problems of time-consuming model training and complex construction of ordinary processors in the above-mentioned prior art, and to provide a deep learning algorithm implementation method and platform based on domestic many-core processors

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
  • Depth learning algorithm achieving method and platform based on domestic many-core processor
  • Depth learning algorithm achieving method and platform based on domestic many-core processor
  • Depth learning algorithm achieving method and platform based on domestic many-core processor

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0039] The present invention will be described in further detail below in conjunction with the accompanying drawings:

[0040] Such as Figure 1 to 2 As shown, the present invention provides a deep learning algorithm implementation method based on domestic many-core processors. The data parallel mode of the deep learning algorithm is completed in a master-slave mode, including the following steps

[0041] S1: Copy the network model into n parts, and the training samples are divided into n parts;

[0042] S2: The i-th network model uses the i-th training sample subset for iterative training;

[0043] S3: Upload the parameter gradient to the parameter server when each iteration is completed;

[0044] S4: Download the parameter set from the parameter server before the start of the next iteration as the initial parameters of this iteration;

[0045] The process of deep learning model training requires a large amount of data processing. This solution adopts a data parallel approach, slicing ...

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 belongs to the field of computer information processing and particularly relates to a depth learning algorithm achieving method and platform based on a domestic many-core processor. A design and an achievement mode of a depth learning algorithm data parallel mode are given out; depth learning algorithm interblock progress grade paralleling and achievement are elaborated; a depth learning algorithm interblock threading grade paralleling and achievement mode is explicated; and finally, a hardware platform building scheme is designed based on an optimization mode and strategy of the domestic many-core platform depth learning algorithm. Thus, hardware resources are sufficiently used; solving time for symmetrical positive definite system of linear equations is greatly reduced; calculation energy consumption is reduced; and cost for computer room construction, management and maintenance is reduced.

Description

Technical field [0001] The invention belongs to the field of computer information processing, and specifically relates to a method and platform for implementing a deep learning algorithm based on a domestic many-core processor. Background technique [0002] The deep learning algorithm is to learn a deep non-linear network structure to achieve complex function approximation, characterize the input data distribution is a representation, and show a powerful ability to learn the essential characteristics of the data set from a few sample sets. The essence of deep learning is to learn more useful features by constructing a machine learning model with multiple hidden layers and massive data training, thereby ultimately improving the accuracy of classification or prediction. [0003] Shenwei's domestic many-core processor is a completely domesticized many-core processor for high-performance parallel computing. Its main core is the computing control core. Its function is similar to that of...

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): G06N3/063G06N3/08
CPCG06N3/063G06N3/08
Inventor 王明清刘姝黄雪董昊
Owner ZHENGZHOU YUNHAI INFORMATION 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