https://github.com/xiaohan2012/sdne-keras
Keras implementation of Structural Deep Network Embedding, KDD 2016
https://github.com/xiaohan2012/sdne-keras
autoencoder keras link-prediction network-embedding visualization
Last synced: about 2 months ago
JSON representation
Keras implementation of Structural Deep Network Embedding, KDD 2016
- Host: GitHub
- URL: https://github.com/xiaohan2012/sdne-keras
- Owner: xiaohan2012
- Created: 2017-12-02T19:43:56.000Z (over 7 years ago)
- Default Branch: master
- Last Pushed: 2018-05-19T15:15:09.000Z (about 7 years ago)
- Last Synced: 2024-05-21T17:23:19.662Z (about 1 year ago)
- Topics: autoencoder, keras, link-prediction, network-embedding, visualization
- Language: Python
- Size: 203 KB
- Stars: 92
- Watchers: 3
- Forks: 31
- Open Issues: 6
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Keras implementation of Structural Deep Network Embedding, KDD 2016
- [The paper](http://www.kdd.org/kdd2016/subtopic/view/structural-deep-network-embedding)
- More details in [my blog article](http://xiaohan2012.github.io/2017/deep-structral-network-embedding-keras/)
# some examples
visualzing 20newsgroup data
- you can also [play with the embedding interactively](http://projector.tensorflow.org/?config=https://gist.githubusercontent.com/xiaohan2012/f9b66b262ba6f92b0f943be896338146/raw/b3cf61184380a435e710d1702a5f84b6fe6896b6/20news-projector-config.json)
label embedding from [stackexchange.datascience](https://datascience.stackexchange.com/) (`deep-learning` as an example)
- you can also [play with the embedding interactively](http://projector.tensorflow.org/?config=https://gist.githubusercontent.com/xiaohan2012/5c533ae2d4c67918c3648a23363307c6/raw/a23dd0b1540b3675d211e5f6db4ffdb969de202d/datascience-tensorboard-config)
# important scripts
## main algorithm
- `core.py`
## experiments
**20newsgroup visualization**
- `20newsgroup_train.py`: train for 20newsgroup dataset
- `20newsgroup_viz.py`: visualization using `sklearn.manifold.TSNE`
- `20newsgroup_tensorboard_embedding.py`: produce the embedding files for [tensorboard projector](https://www.tensorflow.org/versions/r0.12/how_tos/embedding_viz/), which is more interactive
- you can also play with it [here](http://projector.tensorflow.org/?config=https://gist.githubusercontent.com/xiaohan2012/f9b66b262ba6f92b0f943be896338146/raw/b3cf61184380a435e710d1702a5f84b6fe6896b6/20news-projector-config.json) using trained embeddings
**link prediction**
- `link_prediction.py`: train (including grid search) and test
**stackexchange label visualization**
- `stackexchange_train.py`: train for the stackexchange label cooccurence graph
- `stackexchange_label_embedding.py`: produce the embedding files for tensorboard projector
- you can also play with it [here](http://projector.tensorflow.org/?config=https://gist.githubusercontent.com/xiaohan2012/5c533ae2d4c67918c3648a23363307c6/raw/a23dd0b1540b3675d211e5f6db4ffdb969de202d/datascience-tensorboard-config) using trained embeddings
# other implementations
- [suanrong/SDNE](https://github.com/suanrong/SDNE)
- [palash1992/GEM](https://github.com/palash1992/GEM)