Calculation execution method and system for deep neural network

A technology of deep neural network and execution method, which is applied in the direction of biological neural network model, neural architecture, processor architecture/configuration, etc. It can solve problems such as the complexity of application scenarios, efficient and general solutions, and save development and maintenance costs , to achieve the effect of simple and unified, efficient computing resource utilization

Inactive Publication Date: 2017-11-10
CHENDU PINGUO TECH
View PDF7 Cites 14 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In the first direction, many researchers have proposed many effective solutions, but in the second direction, due to the complexity of various specific application scenarios and the difficulty of development, there is no efficient and general solution. solution

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
  • Calculation execution method and system for deep neural network
  • Calculation execution method and system for deep neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0042] In order to make the purpose, technical solution and advantages of the present invention clearer, the present invention will be further elaborated below in conjunction with the accompanying drawings.

[0043] In this example, see figure 1 As shown, the present invention proposes a calculation execution method of a deep neural network, comprising steps:

[0044] S100 builds a deep neural network structure;

[0045] S200 initializes the deep neural network;

[0046] S300 executes the submitted neural layer algorithm through the calculation back-end module, and executes each neural layer module in the deep neural network one by one;

[0047] The S400 obtains the calculation output of the deep neural network by running the calculation back-end module.

[0048] Wherein, the described construction depth neural network structure comprises steps:

[0049] S101 defines all hidden layers in the deep neural network through the neural layer module.

[0050] Define the convolut...

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 discloses a calculation execution method and system for a deep neural network. The calculation execution method comprises the steps of: constructing a deep neural network structure; initializing the deep neural network; by a calculation rear-end module, executing a submitted neural layer algorithm, and executing each neural layer module in the deep neural network one by one; and by operating the calculation rear-end module, acquiring an operation output result of the deep neural network. According to the invention, execution efficiency for a deep neural network model is improved, and calculation resources, such as a GPU, a CPU and the like, can be effectively utilized to a great extent.

Description

technical field [0001] The invention belongs to the technical field of neural networks, and in particular relates to a calculation execution method and system of a deep neural network. Background technique [0002] In recent years, relying on the advancement of deep learning technology, more and more image processing problems can be solved well through deep learning, such as: image recognition, classification and description, pixel-accurate image content segmentation (commonly known as cutout) , image style transfer and so on. However, the calculation and memory overhead of the deep neural network model are very large during execution, and usually can only be run on servers equipped with high-performance GPUs. In order to make deep learning technology more widely applied and landed, it can be used on devices with limited computing resources, including low-end computers, smart phones, smart watches and other mobile devices. [0003] To achieve the above goals, two methods a...

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): G06T1/20G06N3/04G06T1/60
CPCG06N3/04G06T1/20G06T1/60
Inventor 张靖淇
Owner CHENDU PINGUO TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap