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

Traffic data restoration method based on generative adversarial network

A technology of traffic data and repair method, which is applied to data error detection, electrical digital data processing, and response error generation in the direction of redundancy in computing, and can solve problems such as performance degradation

Inactive Publication Date: 2019-07-16
BEIJING UNIV OF TECH
View PDF2 Cites 13 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] However, since these two methods are mainly based on the proximity value and the time series characteristics of traffic data, the performance of these methods will drop significantly when the data changes are severe or the data loss is serious.

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
  • Traffic data restoration method based on generative adversarial network
  • Traffic data restoration method based on generative adversarial network
  • Traffic data restoration method based on generative adversarial network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0033] In the actual production environment, the recording equipment often loses the recording information of some periods, that is, the complete traffic flow matrix (a) in the upper left corner is missing under natural conditions to form the missing matrix (b). We input the missing matrix (b) into the trained generation network to obtain the repair matrix (c), and add the identification network to train the generation network to ensure the accuracy of the repair results. figure 1 middle x ij is the record of complete data, p ij Is the repaired record, ? is the missing record.

[0034] figure 2 Show the specific structure of the trained model. We feed the missing traffic flow matrix into the generative network, first extract features through convolutional and dilated convolutional layers, and then output the inpainting matrix through convolutional and inverse convolutional layers. After the repair, the traffic matrix is ​​judged by the discriminator network to determine ...

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

A traffic data restoration method based on a generative adversarial network is a data restoration method of machine learning. A missing traffic flow matrix often affects the performance of the intelligent traffic system, such as wrong congestion prediction and route guidance. According to the method, the adversarial training idea is introduced into traffic flow data restoration, and the whole model is divided into two parts, namely a generation network and an identification network. A network part repair missing traffic flow matrix is generated, whether the repaired traffic flow matrix conforms to real distribution or not is judged by the identification network, and the time sequence coherence of the traffic flow matrix is ensured by utilizing consistency constraint. According to the method, the accuracy of traffic data restoration is improved, and accurate estimation of the traffic flow incomplete part is realized.

Description

technical field [0001] The invention is a data restoration method of machine learning, and is especially suitable for restoration of missing traffic data. Background technique [0002] Among the matrix inpainting methods, proximity interpolation and inpainting methods based on sparse low-rank matrices are more commonly used methods. [0003] However, since these two methods are mainly based on the proximity value and the time series characteristics of traffic data, the performance of these methods will drop significantly when the data changes are severe or the data loss is serious. [0004] In our invention, we mainly focus on solving the problem that algorithms do not perform well under conditions of severe traffic flow changes or severe losses. On the basis of convolutional network repair, we add an adversarial generative network to improve the effect. As far as we know, we are the first to introduce generative adversarial network into the traffic repair problem. The rep...

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): G06F11/14G06K9/62G06N99/00
CPCG06F11/1402G06F18/214
Inventor 张勇原野王博岳尹宝才
Owner BEIJING UNIV OF TECH
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