This is the repository for the work "An ensemble mechanism to tackle the heterogeneity in asynchronous federated learning"
##requirements
python tensorflow keras
##run
1.[Generate Local Data Sets mnist.py]
Generate non-IID data set saved in [data_mnist_noniid].
2.[mnist_cnn.py]
Generate initialized model saved in [w0.h5] and see local model structure in detail.
3.[FedAvg_mnist.py]
Federated averaging algorithm.
4.[AMU_rdm_mnist.py]
Employ adaptive model updating strategy (AMU).
5.[AMU+TVW+IDW-IE_mnist.py] [AMU+TVW+IDW-LN_mnist.py]
Employ AMU + TVW(time variety weighted strategy) + IDW( information diversity weighted strategy (IDW)).