Secure multi-party computation for personalized human activity recognition
- Author
- David Melanson, Ricardo Maia, Hee-Seok Kim, Anderson Nascimento and Martine De Cock (UGent)
- Organization
- Abstract
- Calibrating Human Activity Recognition (HAR) models to end-users with Transfer Learning (TL) often yields significant accuracy improvements. Such TL is by design done based on very personal data collected by sensors worn close to the human body. To protect the users' privacy, we therefore introduce Secure Multi-Party Computation (MPC) protocols for personalization of HAR models, and for secure activity recognition with the personalized models. Our MPC protocols do not require the end-users to reveal their sensitive data in an unencrypted manner, nor do they require the application developer to disclose their trained model parameters or any other sensitive or proprietary information with anyone in plaintext. Through experiments on HAR benchmark datasets, we demonstrate that our privacy-preserving solution yields the same accuracy gains as TL in-the-clear, i.e. when no measures to protect privacy are in place, and that our approach is fast enough for use in practice.
- Keywords
- Transfer learning, Human activity recognition, Convolutional neural network, Secure multi-party computation, Cryptography, Privacy
Downloads
-
dmelanson2023a.pdf
- full text (Accepted manuscript)
- |
- open access
- |
- |
- 615.59 KB
-
(...).pdf
- full text (Published version)
- |
- UGent only
- |
- |
- 514.05 KB
Citation
Please use this url to cite or link to this publication: http://hdl.handle.net/1854/LU-01HMYSH91YY7MBT6KFQTD5EHZ7
- MLA
- Melanson, David, et al. “Secure Multi-Party Computation for Personalized Human Activity Recognition.” NEURAL PROCESSING LETTERS, vol. 55, no. 3, 2023, pp. 2127–53, doi:10.1007/s11063-023-11182-8.
- APA
- Melanson, D., Maia, R., Kim, H.-S., Nascimento, A., & De Cock, M. (2023). Secure multi-party computation for personalized human activity recognition. NEURAL PROCESSING LETTERS, 55(3), 2127–2153. https://doi.org/10.1007/s11063-023-11182-8
- Chicago author-date
- Melanson, David, Ricardo Maia, Hee-Seok Kim, Anderson Nascimento, and Martine De Cock. 2023. “Secure Multi-Party Computation for Personalized Human Activity Recognition.” NEURAL PROCESSING LETTERS 55 (3): 2127–53. https://doi.org/10.1007/s11063-023-11182-8.
- Chicago author-date (all authors)
- Melanson, David, Ricardo Maia, Hee-Seok Kim, Anderson Nascimento, and Martine De Cock. 2023. “Secure Multi-Party Computation for Personalized Human Activity Recognition.” NEURAL PROCESSING LETTERS 55 (3): 2127–2153. doi:10.1007/s11063-023-11182-8.
- Vancouver
- 1.Melanson D, Maia R, Kim H-S, Nascimento A, De Cock M. Secure multi-party computation for personalized human activity recognition. NEURAL PROCESSING LETTERS. 2023;55(3):2127–53.
- IEEE
- [1]D. Melanson, R. Maia, H.-S. Kim, A. Nascimento, and M. De Cock, “Secure multi-party computation for personalized human activity recognition,” NEURAL PROCESSING LETTERS, vol. 55, no. 3, pp. 2127–2153, 2023.
@article{01HMYSH91YY7MBT6KFQTD5EHZ7,
abstract = {{Calibrating Human Activity Recognition (HAR) models to end-users with Transfer Learning (TL) often yields significant accuracy improvements. Such TL is by design done based on very personal data collected by sensors worn close to the human body. To protect the users' privacy, we therefore introduce Secure Multi-Party Computation (MPC) protocols for personalization of HAR models, and for secure activity recognition with the personalized models. Our MPC protocols do not require the end-users to reveal their sensitive data in an unencrypted manner, nor do they require the application developer to disclose their trained model parameters or any other sensitive or proprietary information with anyone in plaintext. Through experiments on HAR benchmark datasets, we demonstrate that our privacy-preserving solution yields the same accuracy gains as TL in-the-clear, i.e. when no measures to protect privacy are in place, and that our approach is fast enough for use in practice.}},
author = {{Melanson, David and Maia, Ricardo and Kim, Hee-Seok and Nascimento, Anderson and De Cock, Martine}},
issn = {{1370-4621}},
journal = {{NEURAL PROCESSING LETTERS}},
keywords = {{Transfer learning,Human activity recognition,Convolutional neural network,Secure multi-party computation,Cryptography,Privacy}},
language = {{eng}},
number = {{3}},
pages = {{2127--2153}},
title = {{Secure multi-party computation for personalized human activity recognition}},
url = {{http://doi.org/10.1007/s11063-023-11182-8}},
volume = {{55}},
year = {{2023}},
}
- Altmetric
- View in Altmetric
- Web of Science
- Times cited: