Method and device for neural network training based on block chain, and medium

A neural network training and neural network model technology, which is applied in the blockchain field, can solve the problems of difficult and difficult implementation of neural network training, and achieve the effects of cost saving, low compatibility and low difficulty

Inactive Publication Date: 2018-11-27
GUANGDONG UNIV OF TECH
View PDF1 Cites 26 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] However, in order to achieve the effect of "deep learning" after neural network training, it is still difficult to achieve neural network training only with separate computer hardware.
The current proposed neural network training strategy usually adopts the "cloud computing" solution, but "cloud computing" has relatively high requirements for the compatibility between devices, and requires the compatibility of all devices to meet the same standard, so It is difficult to implement neural network training through "cloud computing"

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
  • Method and device for neural network training based on block chain, and medium
  • Method and device for neural network training based on block chain, and medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0043] figure 1 It is a flowchart of a blockchain-based neural network training method provided by an embodiment of the present invention. Please refer to figure 1 , the specific steps of the blockchain-based neural network training method include:

[0044] Step S10: Respond to the training request initiated by the target node, and obtain the neural network model and sample data provided by the target node.

[0045] The main content of this step is to respond to the training request initiated by the neural network. The target node can be a node outside the blockchain network. In this case, it is necessary to communicate between the target node and the nodes in the blockchain network. Control of nodes in the blockchain network. As a more preferred way, the target node can also be any node in the blockchain network. Since the target node and other nodes are under the same network mechanism, the communication efficiency between the target node and other nodes can also be guara...

Embodiment 2

[0058] On the basis of the above embodiments, the present invention also provides the following series of preferred implementation manners.

[0059] As a preferred implementation, before selecting available nodes in the blockchain network that meet preset conditions, the method further includes:

[0060] Obtain the expected training time preset by the target node, and calculate the training time required for each node in the blockchain network to train the neural network model;

[0061] Correspondingly, the preset condition is specifically that the training time is less than the expected training time.

[0062] It should be noted that due to the low performance requirements of the blockchain network for nodes, considering that the performance of nodes in the blockchain network may be uneven, nodes with lower performance cannot undertake the task of more complex neural network models. Therefore, in this embodiment, the target node presets the expected training time, that is, t...

Embodiment 3

[0072] The embodiment of the neural network training method based on the block chain has been described in detail above, and the present invention also provides a neural network training device based on the block chain corresponding to the method, because the embodiment of the device part is the same as The embodiments of the method part correspond to each other, so for the embodiments of the device part, please refer to the description of the embodiments of the method part, and details will not be repeated here.

[0073] figure 2 A structural diagram of a blockchain-based neural network training device provided by an embodiment of the present invention. The block chain-based neural network training device provided by the embodiment of the present invention includes:

[0074] Response acquisition module 10, used to respond to the training request initiated by the target node, and obtain the neural network model and sample data provided by the target node;

[0075] Node sele...

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 method and a device for neural network training based on a block chain, and a medium. The method comprises: responding to a training request initiated by a target node, and obtaining a neural network model and sample data provided by the target node; selecting available nodes which meet preset conditions in a block chain network; dividing the sample data into data segments, and distributing the neural network model and the data segments to each available node; controlling each available node to train the neural network model according to the data segments, and obtaining result models; combining each result model to generate neural network training results. The method ensures effect of neural network training relatively, and difficulty of realizing neural network training is relatively low. In addition, the invention also provides a neural network training device based on a block chain and a medium, and beneficial effects are as described above.

Description

technical field [0001] The present invention relates to the technical field of blockchain, in particular to a neural network training method, device and medium based on blockchain. Background technique [0002] After being proposed by many scientists more than ten years ago, the neural network quickly faded away, and its research and development encountered huge obstacles. The main reason was that it was limited by the limitation of computing resources and the shortage of data sources, which made it impossible to train the neural network. Provide strong support. In recent years, due to the development of integrated circuits, computer hardware can gradually adapt to the high-intensity calculation of neural network training. In addition, due to the development of sensor technology, the data samples used for neural network training are also gradually enriched. Therefore, neural network training has once again become Current technology trends. [0003] However, in order to ach...

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): G06N3/08
CPCG06N3/08
Inventor 陈荣权景东熊晓明
Owner GUANGDONG UNIV OF 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