Skip to content

huaxiuyao/ARML

Repository files navigation

ARML (Automated Relational Meta-learning)

About

Source code1 of the paper Automated Relational Meta-learning

If you find this repository useful in your research, please cite the following paper:

@inproceedings{
  yao2020automated,
  title={Automated Relational Meta-learning},
  author={Huaxiu Yao and Xian Wu and Zhiqiang Tao and Yaliang Li and Bolin Ding and Ruirui Li and Zhenhui Li},
  booktitle={International Conference on Learning Representations},
  year={2020},
  url={https://openreview.net/forum?id=rklp93EtwH}
}

Data

Plain-Multi and Art-Multi are released in Google Drive

Usage

Dependence

  • python 3.*
  • TensorFlow 1.10+
  • Numpy 1.15+

2-D Regression

Please see the bash file in /2D_bash for hyperparameter settings

Plainmulti

Please see the bash file in /plainmulti_bash for hyperparameter settings

Artmulti

Please see the bash file in /artmulti_bash for hyperparameter settings

1This code is built based on the MAML and HSML.

About

ARML for ICLR 2020

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published