Differential privacy in statistical queries for synthetic trajectories generated by generative adversarial networks.
Saved in:
| Title: | Differential privacy in statistical queries for synthetic trajectories generated by generative adversarial networks. |
|---|---|
| Authors: | Shin, Jihwan (AUTHOR), Song, Yeji (AUTHOR), Jang, Minsoo (AUTHOR), Ahn, Jinhyun (AUTHOR), Lee, Taewhi (AUTHOR), Im, Dong-Hyuk (AUTHOR) |
| Source: | Connection Science. Dec 2025, Vol. 37 Issue 1, p1-21. 21p. |
| Subjects: | Privacy, Data privacy, Statistics, Anonymity, Location-based services, Generative adversarial networks, Data management, Trajectories (Mechanics) |
| Abstract: | With the widespread adoption of smartphones and the rapid advancement of information and communication technologies, the use of Location-Based Services (LBS) has significantly increased across various domains. Consequently, the collection and utilisation of user trajectory data are also growing rapidly. While such data can provide valuable insights for personalised services and other analyses, it inherently contains sensitive location information, posing serious privacy risks if used without proper anonymization. Previous studies have attempted to mitigate privacy concerns by applying Differential Privacy (DP) to prefix tree structures for statistical analysis. However, these approaches often suffer from diminished data utility due to the excessive noise required by DP mechanisms. To address this issue, we propose a two-stage trajectory privacy framework. In the first stage, we employ a Category Auxiliary Classifier-Generative Adversarial Network (CAC-GAN) to generate synthetic trajectory data that preserves the statistical characteristics of the original data, thereby providing primary privacy protection. In the second stage, we apply a prefix tree-based DP algorithm to the synthetic data, offering enhanced privacy during statistical analysis and query processing. Experimental results demonstrate that the proposed CAC-GAN method achieves approximately 53% improvement in both data utility and anonymity compared to existing methods. Furthermore, relative error analysis across various ϵ values confirms that our two-stage protection scheme maintains superior statistical accuracy. This study presents a novel methodology that effectively balances trajectory data privacy and utility. [ABSTRACT FROM AUTHOR] |
| Copyright of Connection Science is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Psychology and Behavioral Sciences Collection |
|
Full text is not displayed to guests.
Login for full access.
|
|
| Abstract: | With the widespread adoption of smartphones and the rapid advancement of information and communication technologies, the use of Location-Based Services (LBS) has significantly increased across various domains. Consequently, the collection and utilisation of user trajectory data are also growing rapidly. While such data can provide valuable insights for personalised services and other analyses, it inherently contains sensitive location information, posing serious privacy risks if used without proper anonymization. Previous studies have attempted to mitigate privacy concerns by applying Differential Privacy (DP) to prefix tree structures for statistical analysis. However, these approaches often suffer from diminished data utility due to the excessive noise required by DP mechanisms. To address this issue, we propose a two-stage trajectory privacy framework. In the first stage, we employ a Category Auxiliary Classifier-Generative Adversarial Network (CAC-GAN) to generate synthetic trajectory data that preserves the statistical characteristics of the original data, thereby providing primary privacy protection. In the second stage, we apply a prefix tree-based DP algorithm to the synthetic data, offering enhanced privacy during statistical analysis and query processing. Experimental results demonstrate that the proposed CAC-GAN method achieves approximately 53% improvement in both data utility and anonymity compared to existing methods. Furthermore, relative error analysis across various ϵ values confirms that our two-stage protection scheme maintains superior statistical accuracy. This study presents a novel methodology that effectively balances trajectory data privacy and utility. [ABSTRACT FROM AUTHOR] |
|---|---|
| ISSN: | 09540091 |
| DOI: | 10.1080/09540091.2025.2523964 |
