Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

A training method for automatic operation and maintenance strategy of information system

A technology of information system and training method, which is applied in software deployment and other directions, and can solve problems such as system crashes

Active Publication Date: 2020-10-16
POLIXIR TECH LTD
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the learning process of reinforcement learning requires a lot of trial-and-error interaction between the agent and the environment. In the actual automatic operation and maintenance problem solving, direct use of reinforcement learning requires the use of a large number of different parameter configurations for trial and error, which may lead to the collapse of the existing system. , it is obviously impractical to use reinforcement learning methods directly in real information systems

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
  • A training method for automatic operation and maintenance strategy of information system
  • A training method for automatic operation and maintenance strategy of information system
  • A training method for automatic operation and maintenance strategy of information system

Examples

Experimental program
Comparison scheme
Effect test

approach example 1

[0021] [Scheme Example 1] Using supervised learning

[0022] First, several neural networks are established as models of user behavior strategies and information system background.

[0023] Secondly, organize the real information system data into a data set D={(u,c,q,a,u')}, each piece of data corresponds to a tuple (u,c,q,a,u'), where u Indicates the current user set, c indicates the configured system parameters, q indicates the corresponding service quality under parameter c, a indicates user behavior, and u' indicates the new user set after the user performs action a.

[0024] Then, using the real (u, c) in the real data set D as input and the real service quality q as output, a supervised learning method is used to train the service quality evaluation function.

[0025] Again, use the real (u, q) in the real data set D as the input, and the real user behavior a as the output, and use the supervised learning method to train the user behavior strategy.

[0026] Then, use t...

Embodiment 1

[0040] [Example 1] Joint training (using confrontational learning and reinforcement learning)

[0041] First, several neural networks are established as models of user behavior strategies and information system background.

[0042] Secondly, establish a neural network as a discriminator to judge the credibility of the data. The value of the credibility is a real number between 0 and 1. The closer to 1, the more it looks like real data, and the closer it is to 0, the more it looks like generated data. .

[0043] Again, in the step 3)-6) of the information system background and user behavior strategy training process, the joint output of the user behavior strategy and the information system background is spliced ​​into a tuple (u,c,q,a,u') , execute step 7), and construct a simulated data set D'={(u,c,q,a,u')}, use the real data set D and the simulated data set D' during training as data, and update a discriminant device, the update target is as follows:

[0044]

[0045] ...

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 training method for an automatic operation and maintenance strategy of an information system. The method mainly comprises three parts of information system simulator construction based on machine learning and adversarial learning, automatic intelligent operation and maintenance strategy search based on reinforcement learning and automatic intelligent operation and maintenance strategy model migration optimization, and the problems of high dynamic and difficult solution in the field of information systems are solved.

Description

technical field [0001] The invention relates to a training method for an automatic operation and maintenance strategy of an information system, which can be used for automatic and intelligent operation and maintenance of an information system, and belongs to the technical field of intelligent operation and maintenance. Background technique [0002] The goal of the automatic intelligent operation and maintenance strategy is to assist the information system to adjust and maintain the parameters of each device node in the system according to the real-time system status, and provide the best information service for users within the coverage area in real time. The traditional automatic operation and maintenance strategy design mostly relies on making certain preference assumptions or constraints on surrounding users, and then using traditional optimization methods to solve them. These assumptions are usually difficult to accurately describe the behavior of surrounding users, and t...

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 Patents(China)
IPC IPC(8): G06F8/60
CPCG06F8/60
Inventor 俞扬秦熔均
Owner POLIXIR TECH LTD
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
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