Skip to content

sustainable-computing/Blinder

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Blinder: End-to-end Privacy Protection in Sensing Systems via Personalized Federated Learning

arXiv License

This repository contains the implementation of the paper entitled "Blinder: End-to-end Privacy Protection in Sensing Systems via Personalized Federated Learning".

Directories:

Datasets

Blinder is evaluated on two Human Activity Recognition (HAR) datasets: MotionSense and MobiAct.

The datasets and the preprocessing script (required for MobiAct) are available at:

Note: Pre-trained evaluation models can be found under Blinder-Python/eval_models/.

Dependencies

Package Version
Python3 3.8.13
PyTorch 1.10.2
TensorFlow 2.8.0
imbalanced_learn 0.9.0
scikit-learn 1.1.2

Installation

Android App:

Blinder is deployed on Android platforms for real-time data anonymization. This deployment uses Blinder models pre-trained on MobiAct and MotionSense.

Acknowledgement

  • learn2learn: a software library for meta-learning research.

Citation

Xin Yang and Omid Ardakanian. 2023. Blinder: End-to-end Privacy Protection in Sensing Systems via Personalized Federated Learning. ACM Trans. Sen. Netw. 20, 1, Article 15 (January 2024), 32 pages.

@article{yang2023blinder,
  title={Blinder: End-to-end Privacy Protection in Sensing Systems via Personalized Federated Learning},
  author={Yang, Xin and Ardakanian, Omid},
  journal={ACM Transactions on Sensor Networks},
  volume={20},
  number={1},
  pages={1--32},
  year={2023},
  publisher={ACM New York, NY}
}