The original authors are
- Kang Wei
- Jun Li
- Chuan Ma
- Ming Ding
- Sha Wei
- Fan Wu
- Guihai Chen
- Thilina Ranbaduge
Corresponding paper can be found here: https://arxiv.org/abs/2202.04309
Original github repository can be found here: https://github.com/AdamWei-boop/Vertical_FL
There were some errors which probably stem from not having the exact same data. The data was not included in the original zip, so I added it myself to the best of my ability.
Finally I was able to reproduce everything with only minor changes to the code.
Python 3.8
Torch 1.8.0 or Tensorflow 2.3.0
pip3 insatll pipenv
pipenv install -r requirements.txt
pipenv shell
Learning models: neural networks
Datasets: Adult
parser.add_argument('--dname', default='ADULT', help='dataset name')
parser.add_argument('--epochs', type=int, default=10, help='number of training epochs')
parser.add_argument('--batch_type', type=str, default='mini-batch')
parser.add_argument('--batch_size', type=int, default=512)
parser.add_argument('--data_type', default='original', help='define the data options: original or one-hot encoded')
parser.add_argument('--model_type', default='vertical', help='define the learning methods: vrtical or centralized')
parser.add_argument('--organization_num', type=int, default='3', help='number of origanizations, if we use vertical FL')
python tf_vertical_FL_train.py --epochs 100 --model_type 'vertical' --organization_num 3
python torch_vertical_FL_train.py --epochs 100 --model_type 'vertical' --organization_num 3
- Ashely make sure the code can work and know the execution flow
- Adrian make sure the code can work and know the execution flow
- Adrain upload the Differential privacy into your branch please