Awesome

HSML (Hierarchically Structured Meta-learning)

About

Source code<a href="#note1" id="note1ref"><sup>1</sup></a> of the paper Hierarchically Structured Meta-learning

For continual version of this algorithm, please refer to this <a href="#note1" id="note1ref"></a> repo.

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

@inproceedings{yao2019hierarchically,
  title={Hierarchically Structured Meta-learning},
  author={Yao, Huaxiu and Wei, Ying and Huang, Junzhou and Li, Zhenhui},
  booktitle={Proceedings of the 36th International Conference on Machine Learning},
  year={2019} 
}

Data

We release our Multi-Datasets including bird, texture, aircraft and fungi in this link.

Usage

Dependence

python 3.*
TensorFlow 1.0+
Numpy 1.15+

Toy Group Data

Please see the bash file in /toygroup_bash for parameter settings

Multi-datasets Data

Please see the bash file in /multidataset_bash for parameter settings

<a id="note1" href="#note1ref"><sup>1</sup></a>This code is built based on the MAML.