AutoNE

The Implementation of "AutoNE: Hyperparameter Optimization for Massive Network Embedding"(KDD 2019).

Requirements

• Python 3

$ pip3 install -r requirements.txt

Besides, this code relies on my submodule embeddding_test. Please use following command to download the code:

$ git clone --recursive git@github.com:tadpole/AutoNE.git

Usage

The Dataset can be downloaded from here.

You can change 'dataset', 'method', 'task', 'ms' variables in Makefile to select data and model.

dataset :   [BlogCatalog | Wikipedia | pubmed]
method  :   [deepwalk | AROPE | gcn]
task    :   [link_predict | classification]
ms      :   [mle | random_search | b_opt]

Sampling dataset

$ make sample

Run the model

$ make run

Cite

If you find this code useful, please cite our paper:

@inproceedings{tu2019autone,
  title={AutoNE: Hyperparameter Optimization for Massive Network Embedding},
  author={Tu, Ke and Ma, Jianxin and Cui, Peng and Pei, Jian and Zhu, Wenwu},
  booktitle={Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery \& Data Mining},
  year={2019},
  organization={ACM}
}