Qualifying Exam

Qualifying Exam

Urban-scale Human Mobility Data Synthesis via Learning-based models with Privacy Awareness


Download as iCal file

Wednesday, November 04, 2020, 03:30pm - 05:00pm


Speaker: Fan Zhang

Location : Remote via Webex


Prof. Desheng Zhang (Advisor)

Prof. Jie Gao

Prof. Yongfeng Zhang

Prof. Eric Allender

Event Type: Qualifying Exam

Abstract: Real-time human mobility data obtained from urban infrastructure (e.g., cellular networks) have the potential to unleash the full power of social and collaborative computing systems and applications, from crowd sourcing, to ridesharing, mobile networking, trans- portation planning, emergency response, and pandemic mitigation. However, most of the city-wide human mobility data are often pro- prietary and thus cannot be accessed by the research community (e.g., researchers and practitioners) unless released by data owners (e.g., government or industries). Recently, under the context of Data Science for Social Good, some data owners are willing to share their mobility data with the public to unlock their value, but there are significant privacy and security concerns that remain in the way. To overcome these issues, we present a privacy-aware framework to generate synthetic yet realistic mobility data through augmented Generative Adversarial Nets (GANs) based on real world human mobility data from a cellular network. We plug in two auxiliary modules, i.e., privacy classifier and utility classifier, to apply targeted perturbation, which balances the trade-off between sensitive and non-sensitive mobility features. We implement our framework to generate urban-scale cellular connection records based on large scale real world data from Shenzhen cellular networks as a case study and empirically evaluate its performance in both utility and privacy aspects. The quantitative results show that with same level of privacy risks, our model outperforms the baseline models by 33% gain on temporal utility and an up to 49% gain on spatiotemporal utility.


Wednesday, Nov 4, 2020 3:30 pm | 1 hour 30 minutes | (UTC-04:00) Eastern Time (US & Canada)
Meeting number: 120 492 9656
Password: gE3niQxxp23

Join by video system
Dial This email address is being protected from spambots. You need JavaScript enabled to view it.
You can also dial and enter your meeting number.

Join by phone
+1 650-429-3300 USA Toll
Access code: 120 492 9656