PPML-TSA: A modular privacy-preserving time series classification framework

• DOI: 10.1016/j.simpa.2022.100286
• Publication Date: 2022-04-09T01:41:52Z
• Authors: Dominique Mercier; Adriano Lucieri; Mohsin Munir; Andreas Dengel; Sheraz Ahmed
• Keywords: Differential Privacy; Homomorphic Encryption; Implementation

Privacy-preservation is of key importance for the transition modern deep learning algorithms into everyday applications dealing with sensitive data, such as healthcare, finance and several other domains critical infrastructure. One major impediment research in computer science considerable time investment required to set up experiments their evaluation. In domain privacy-preserving learning, this aggravated by dispersion implementations throughout frameworks libraries. This work introduces documents PPML-TSA, a versatile framework series classification. Our was initially used evaluate methods across different model architectures datasets. PPML-TSA offers modular design suitable performing classification on all datasets from entire UCR & UEA repository. Its implementation quick easy adaptation extension increase number supported The code supports variety (such AlexNet, FCN, FDN, LSTM, LeNet) well (Differential Privacy, Federated Learning, combination Homomorphic Encryption) out-of-the-box. We believe that our facilitates further resulting accelerated innovation disruption field.