EAML

Code release for "Learning to Adapt to Evolving Domains" (NeurIPS 2020)

Prerequisites

• PyTorch >= 0.4.0 (with suitable CUDA and CuDNN version)
• torchvision >= 0.2.1
• Python3
• Numpy
• argparse
• PIL

Dataset

Rotated MNIST: https://drive.google.com/file/d/1eaw42sg4Cgm34790AW_SKGCSkFosugl2/view?usp=sharing

Training

EAML 

%run eaml.py rot_mnist_28/ --lip-balance 0.2 --lip-jth 0.01 --epochs 500 --lr-in 0.03 --lr-out 0.003 

JAN 

%run JAN.py rot_mnist_28/ --lip-balance 0.2 --lip-jth 0.01 --epochs 500 --lr-in 0.03 --lr-out 0.003

Source 

%run source.py rot_mnist_28/ --lip-balance 0.2 --lip-jth 0.01 --epochs 500 --lr-out 0.003

Acknowledgement

This code is implemented based on the JAN (Joint Adaptation Networks) code, and it is our pleasure to acknowledge their contributions. The meta-learning code is adapted from https://github.com/dragen1860/MAML-Pytorch/.

Citation

If you use this code for your research, please consider citing:

@inproceedings{NEURIPS2020_fd69dbe2,
 author = {Liu, Hong and Long, Mingsheng and Wang, Jianmin and Wang, Yu},
 booktitle = {Advances in Neural Information Processing Systems},
 editor = {H. Larochelle and M. Ranzato and R. Hadsell and M. F. Balcan and H. Lin},
 pages = {22338--22348},
 publisher = {Curran Associates, Inc.},
 title = {Learning to Adapt to Evolving Domains},
 url = {https://proceedings.neurips.cc/paper/2020/file/fd69dbe29f156a7ef876a40a94f65599-Paper.pdf},
 volume = {33},
 year = {2020}
}

Contact

If you have any problem about our code, feel free to contact

• [email protected]