https://github.com/abhilash1910/deep-graph-learning
A notebook containing implementations of different graph deep node embeddings along with benchmark graph neural network models in tensorflow. This has been taken from https://www.kaggle.com/abhilash1910/nlp-workshop-ml-india-deep-graph-learning to apply GNNs/node embeddings on NLP task.
https://github.com/abhilash1910/deep-graph-learning
chebyshev-filter gcn gnn graphneuralnetwork keras-tensorflow laplacian-filter line sdne semi tensorflow2
Last synced: 3 months ago
JSON representation
A notebook containing implementations of different graph deep node embeddings along with benchmark graph neural network models in tensorflow. This has been taken from https://www.kaggle.com/abhilash1910/nlp-workshop-ml-india-deep-graph-learning to apply GNNs/node embeddings on NLP task.
- Host: GitHub
- URL: https://github.com/abhilash1910/deep-graph-learning
- Owner: abhilash1910
- Created: 2021-07-16T19:29:40.000Z (about 4 years ago)
- Default Branch: master
- Last Pushed: 2021-07-17T06:21:38.000Z (about 4 years ago)
- Last Synced: 2025-06-30T01:18:03.218Z (3 months ago)
- Topics: chebyshev-filter, gcn, gnn, graphneuralnetwork, keras-tensorflow, laplacian-filter, line, sdne, semi, tensorflow2
- Language: Jupyter Notebook
- Homepage: https://www.kaggle.com/abhilash1910/nlp-workshop-ml-india-deep-graph-learning
- Size: 3.29 MB
- Stars: 13
- Watchers: 1
- Forks: 3
- Open Issues: 0
-
Metadata Files:
- Readme: Readme.md
Awesome Lists containing this project
README
## Deep Graph Learning
This [notebook](https://www.kaggle.com/abhilash1910/nlp-workshop-ml-india-deep-graph-learning) contains the implementations of different deep learning algorithms applied on graphs.This includes implementations of SDNE and LINE, which fall under deep node embeddings. Both these embedding models are built in Tensorflow and rely on 2 order neighbors for generating the node representations. The intrinsic model is a simple linear Dense model with softmax activation to generate the embedding space.Post that there is an indepth implementation of Graph Neural Network architectures particularly suited for classifying node embedding representation from any networkx graph.This follows Thimas Kpif's vanilla spectral GCN, along with other variations of Spectral GNN including Laplacian GCN, Spline GCN and ChebNets. The implementations are also in Tensorflow /Keras.The following are the contents of the notebook:
## Deep Node Embeddings
### SDNE (Structural Deep Network Embeddings)
Resources
- [Paper](http://www.kdd.org/kdd2016/papers/files/rfp0191-wangAemb.pdf)
- [Github](https://github.com/suanrong/SDNE)
### LINE (Large Scale Information Network Embedding)
Resources
- [Paper](https://arxiv.org/abs/1503.03578v1)
- [Github](https://github.com/tangjianpku/LINE)
## Spectral Graph Neural
### Vanilla GNN
Resources
- [Paper](https://arxiv.org/abs/1609.02907)
- [Github](https://github.com/tkipf/gcn)
### Spline GCN
Resources
- [Paper](https://openaccess.thecvf.com/content_cvpr_2018/papers/Fey_SplineCNN_Fast_Geometric_CVPR_2018_paper.pdf)
- [Code](https://paperswithcode.com/paper/splinecnn-fast-geometric-deep-learning-with/review/)
### Laplacian GCN
- [Paper](https://arxiv.org/abs/1809.09839)
### ChebNets GCN
- [Paper](https://arxiv.org/abs/1911.05467)
- [Code](https://github.com/rusty1s/pytorch_geometric)