This repository consists of implementation of secure federated learning system that uses encrypted weight exchanges for secure exchange of weights and anomaly detection to identify byzantine client. FedAvg is used for aggregation of weights.
Implementation of PEN model proposed by Xiao et al. (https://www.sciencedirect.com/science/article/abs/pii/S0950705121006006) is provided.
Experiment details and results can be found here: https://drive.google.com/file/d/1XoVht49GHKYr1Rza0FgTH3SHJHvrLpLE/view?usp=sharing