qmkakaxi / DIG_FL

DIGFL: Efficient Participant Contribution Evaluation for Horizontal and Vertical Federated Learning

DIGFL is an efficient participant contribution evaluation for horizontal and vertical Federated Learning.

Here, we persent two versions: standaloneBeta and clusterBeta.

standaloneBeta.

StandaloneBeta is a stand-alone version, to compare the DIGFL with actual shapley value.

Please refer to standaloneBeta.

clusterBeta

ClusterBeta is used to deploy in real FL systems, and can perform federated learning training and contribution calculations on multiple machines or edge devices.

Please refer to clusterBeta.

If you want the MOTOR and REAL dataset, please contact us.