UMDataScienceLab/Personalized_FL_with_DA

The implementation of the paper domain adaptation for personalized federated learning

pflda

The numerical experiment code of the paper Personalized Federated Learning via Domain Adaptation with an Application to Distributed 3D Printing.

1. Sine regression

sine_regression folder contains all the code needed for reproducing results in section 4.1

for instance, try

python3 sineregression.py --seed=6 --fed='PDA'

The fed argument specifies which algorithm to use, three options are 'PDA', 'ditto', and 'indiv'.

All results will be generated to sine_outputs folder

2. Case study on image classification

case_study folder contains all the code needed for reproducing results in section 4.2. Some classes are inherited from the Domainbed repository.

3. 3D printing

printer_dataset folder contains all the code needed for reproducing results in section 4.3