Skip to content

Latest commit

 

History

History
30 lines (18 loc) · 841 Bytes

README.md

File metadata and controls

30 lines (18 loc) · 841 Bytes

Cross-Silo Federated Learning across Divergent Domains with Iterative Parameter Alignment

Published at IEEE International Conference on Big Data 2023

alt text

To get started, check out the basic MNIST examples:

mnist_examples/mnist_mlp_2_.py 
mnist_examples/mnist_mlp_2_detach.py

The following code shows how to run an experiment with a specific configuration. You can specify arguments in the config file or pass them in the python command.

python -u main.py --config=configs/cifar10_3_split_label_Conv4_5.yaml --gpu=2  --weight_seed=32 --seed=32   --same_initialization=True --random_topology=False --local_epochs=3 --merge_iter=3000

Additional examples are in the run_hist file.

Important pieces of code are located in the following files:

utils/align_util.py

models/layers.py