RECT (python source code)

Network Embedding with Completely-imbalanced Labels. TKDE2020 paper. This is a deep method for the problem of Zero-shot Graph Embedding (ZGE), i.e., graph embeddings when labeled data cannot cover all classes.

Breifly, RECT contains two parts:

• RECT-L is the supervised part in which a semantic loss is used.
• RECT-N is the unsupervised part in which the network structure is preserved. Note, this part can be replaced by any unsupervised NRL methods.

Usage (abstract):

• set the dataset
• python main_rect.py

------ evaluate RECT-N ---------
Training an SVM classifier under the pre-defined split setting...
(0.7335058214747736, 0.670830503861163)
------ evaluate RECT-L ---------
Training an SVM classifier under the pre-defined split setting...
(0.7141871496334627, 0.6402691559469643)
------ evaluate RECT ---------
Training an SVM classifier under the pre-defined split setting...
(0.7441138421733506, 0.6805281849343917)

Citing

If you find this useful in your research, please cit our paper, thx:

@article{wang2020RECT,
  title={Network Embedding with Completely-imbalanced Labels},
  author={Wang, Zheng and Ye, Xiaojun and Wang, Chaokun and Cui, Jian and Yu, Philip S},
  journal={TKDE},
  year={2020},
  doi = {10.1109/TKDE.2020.2971490},
  publisher={IEEE}
}