This package provides implementations of Multi-EPL, which is submitted at PLOS ONE.
MultiEPL
├── data
│ └── digits: Digits-Five dataset
├── script
│ └── digits.sh: Shell file for Digits-Five experiments
└── src
├── loader
│ ├── digits: for Digits-Five dataset setting
│ │ ├── mnist.py
│ │ ├── mnist_m.py
│ │ ├── svhn.py
│ │ ├── synthdigits.py
│ │ └── usps.py
│ └── dataloader.py: generate dataloader and dataset of Digits-Five
├── utils
│ └── default_param.py: set default parameters
├── network
│ └── network_digits.py: network for Multi-EPL with Digits-Five dataset
├── solver.py: solver class for training
└── digits.py: code for training Multi-EPL with Digits-Five dataset
- Digits-Five: Consists of five datasets for digit recognition
- MNIST
- MNIST-M
- SVHN
- SynthDigits
- USPS
- Download link: [data]
- Add the data files at "Multi-EPL/data/digits" before running the codes
- Data source: https://github.com/VisionLearningGroup/VisionLearningGroup.github.io/tree/master/M3SDA/code_MSDA_digit
- Note that we are unrelated to the group providing the data.
- Ubuntu
- CUDA 10.0
- Python 3.6.12
- torch 1.7.1
- torchvision 0.8.2
- scipy 1.5.4
To create an anaconda environment with all the requirements:
conda env create -n <ENV_NAME> -f requirement.txtgit clone https://github.com/snudatalab/AUBER.git
cd Multi-EPL/script/
sh digits.sh- Seongmin Lee (ligi214@snu.ac.kr)
- Hyunsik Jeon (jeon185@gmail.com)
- U Kang (ukang@snu.ac.kr)
- Data Mining Lab at Seoul National University.
If you use this code, please cite the following paper.
@article{lee2021multi,
title={Multi-EPL: Accurate multi-source domain adaptation},
author={Lee, Seongmin and Jeon, Hyunsik and Kang, U},
journal={PloS one},
volume={16},
number={8},
pages={e0255754},
year={2021},
publisher={Public Library of Science San Francisco, CA USA}
}