An experimental comparison of mobility learning models for privacy

Type: Bachelor
Status: Assigned March 2021
Student: Roland Holenstein

Human mobility can be modeled using smartphone data, however, such data represents sensitive information, and the collection of those threatens the privacy of the users involved. The recent introduction of federated learning, a privacy-preserving approach to build machine and deep learning models, represents a promising technique to solve the privacy issue.

In the context of this BSc project, you will:

1. Create a summary of recent articles [1], [2], updated with recent work related to the topic.
2. Pre-process a dataset for human mobility modeling (e.g., Foursquare from [3])
3. Create an implementation of one baseline method (e.g., a fully connected network) and one state-of-the-art (existing) deep learning method (e.g., [3])
4. Perform an experimental comparison of at least two methods using the pre-processed dataset
5. Devise an extension of the implemented method to use Federated Learning;
6. Perform an experimental comparison of the methods with and without Federated Learning;
7. Ensure you provide well-documented code.

References:

1. Luca, Massimiliano, Gianni Barlacchi, Bruno Lepri, and Luca Pappalardo. “Deep Learning for Human Mobility: a Survey on Data and Models.” arXiv preprint arXiv:2012.02825 (2020). [https://arxiv.org/pdf/2012.02825.pdf]
2. Yuan, Haitao, and Guoliang Li. “A Survey of Traffic Prediction: from Spatio-Temporal Data to Intelligent Transportation.” Data Science and Engineering (2021): 1-23. [https://link.springer.com/article/10.1007/s41019-020-00151-z]
3. Yang, Dingqi, Benjamin Fankhauser, Paolo Rosso, and Philippe Cudre-Mauroux. “Location Prediction over Sparse User Mobility Traces Using RNNs: Flashback in Hidden States.” In
   Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence, IJCAI20, pp. 2184-2190. 2020.

For more information contact: