Privacy and Robustness in Federated Learning for Healthcare

: The deployment of Federated Learning (FL) in sensitive domains, particularly healthcare, faces significant challenges related to data privacy and model robustness against malicious attacks. Traditional centralized training is infeasible due to strict regulations and the non-IID nature of hospital data. This paper addresses the critical trade-off between privacy preservation and robustness in FL when faced with data poisoning attacks. The primary research question is: Which privacy-preserving techniques are most effective for federated learning in healthcare when facing data poisoning attacks?