Federated Unlearning: The right to be forgotten

Description:  In the era of big data and distributed computing, maintaining privacy and ensuring the right to be forgotten has become increasingly challenging. Federated Unlearning addresses this by developing methodologies to remove specific data points or user data from machine learning models in a federated learning environment without the need to retrain the entire model from scratch. This project aims to create efficient, scalable algorithms and protocols that can perform unlearning tasks across distributed nodes while preserving model accuracy and privacy.

Tasks:

1- Algorithm Development: Design and implement a novel algorithm for federated unlearning that is computationally efficient and can be seamlessly integrated into existing federated learning frameworks.

2- Privacy Preservation: Ensure that the unlearning process maintains user's privacy and adheres to legal requirements such as GDPR.

3- Performance Evaluation: Assess the impact of federated unlearning on model accuracy and performance, providing metrics and benchmarks for effectiveness.

References:

Wu, L., Guo, S., Wang, J., Hong, Z., Zhang, J., & Ding, Y. (2022). Federated unlearning: Guarantee the right of clients to forget. IEEE Network, 36(5), 129-135

Halimi, A., Kadhe, S., Rawat, A., & Baracaldo, N. (2022). Federated unlearning: How to efficiently erase a client in fl?. arXiv preprint arXiv:2207.05521.

Liu, G., Ma, X., Yang, Y., Wang, C., & Liu, J. (2020). Federated unlearning. arXiv preprint arXiv:2012.13891.

Wang, J., Guo, S., Xie, X., & Qi, H. (2022, April). Federated unlearning via class-discriminative pruning. In Proceedings of the ACM Web Conference 2022 (pp. 622-632).

Wang, F., Li, B., & Li, B. (2023). Federated unlearning and its privacy threats. IEEE Network.