Skip to main navigation Skip to search Skip to main content

Traj-MergeGAN: A Trajectory Privacy Preservation Model Based on Generative Adversarial Network

  • Lida Guo
  • , Zimeng Li
  • , Jingyuan Wang*
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Nowadays, with the rapid development of location-based services, individual trajectory data is collected for various traffic related applications. However, while we are benefiting from these services, the trajectory data may contain lots of private information and privacy issues need to be carefully handled. In this paper, we propose a deep learning model named Traj-MergeGAN, which can generate synthetic trajectory from original trajectory. The generated trajectory can not only protect individual privacy, but also maintain data quality for other downstream applications. Furthermore, we conduct overall experiments to prove the advantages of our model.

Original languageEnglish
Title of host publicationKnowledge Science, Engineering and Management - 17th International Conference, KSEM 2024, Proceedings
EditorsCungeng Cao, Huajun Chen, Liang Zhao, Junaid Arshad, Yonghao Wang, Taufiq Asyhari
PublisherSpringer Science and Business Media Deutschland GmbH
Pages361-372
Number of pages12
ISBN (Print)9789819754977
DOIs
StatePublished - 2024
Event17th International Conference on Knowledge Science, Engineering and Management, KSEM 2024 - Birmingham, United Kingdom
Duration: 16 Aug 202418 Aug 2024

Publication series

NameLecture Notes in Computer Science
Volume14886 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference17th International Conference on Knowledge Science, Engineering and Management, KSEM 2024
Country/TerritoryUnited Kingdom
CityBirmingham
Period16/08/2418/08/24

Keywords

  • Deep Learning
  • Generative Adversarial Network
  • Trajectory Privacy

Fingerprint

Dive into the research topics of 'Traj-MergeGAN: A Trajectory Privacy Preservation Model Based on Generative Adversarial Network'. Together they form a unique fingerprint.

Cite this