Differentially private multivariate time series forecasting of aggregated human mobility with deep learning: Input or gradient perturbation?

• Published in: Neural computing & applications Vol. 34; no. 16; pp. 13355 - 13369
• Main Authors: Arcolezi, Héber Hwang, Couchot, Jean-François, Renaud, Denis, Al Bouna, Bechara, Xiao, Xiaokui
• Format: Journal Article
• Language: English
• Published: London Springer London 01.08.2022; Springer Nature B.V; Springer Verlag
• Subjects: Algorithms; Artificial Intelligence; Computational Biology/Bioinformatics; Computational Science and Engineering; Computer Science; Cryptography and Security; Data Mining and Knowledge Discovery; Decision making; Deep learning; Forecasting; Image Processing and Computer Vision; Machine Learning; Mathematical models; Mobility; Multivariate analysis; Perturbation; Privacy; Probability and Statistics in Computer Science; Recurrent neural networks; S.I.: Deep Learning for Time Series Data; Special Issue on Deep Learning for Time Series Data; Time series; Urban planning; Crowd flow; Differential privacy; Differentially private machine learning; Mobility prediction
• ISSN: 0941-0643; 1433-3058; 1433-3058
• DOI: 10.1007/s00521-022-07393-0

This paper investigates the problem of forecasting multivariate aggregated human mobility while preserving the privacy of the individuals concerned. Differential privacy, a state-of-the-art formal notion, has been used as the privacy guarantee in two different and independent steps when training deep learning models. On one hand, we considered gradient perturbation , which uses the differentially private stochastic gradient descent algorithm to guarantee the privacy of each time series sample in the learning stage. On the other hand, we considered input perturbation , which adds differential privacy guarantees in each sample of the series before applying any learning. We compared four state-of-the-art recurrent neural networks: Long Short-Term Memory, Gated Recurrent Unit, and their Bidirectional architectures, i.e., Bidirectional-LSTM and Bidirectional-GRU. Extensive experiments were conducted with a real-world multivariate mobility dataset, which we published openly along with this paper. As shown in the results, differentially private deep learning models trained under gradient or input perturbation achieve nearly the same performance as non-private deep learning models, with loss in performance varying between 0.57 and 2.8 % . The contribution of this paper is significant for those involved in urban planning and decision-making, providing a solution to the human mobility multivariate forecast problem through differentially private deep learning models.