This is a Theano implementation of DGCN, a Dual Graph Convolutional Networks method for graph-based semi-supervised classification proposed in the following paper:
Dual Graph Convolutional Networks for Graph-Based Semi-Supervised Classification. Chenyi Zhuang, Qiang Ma. WWW 2018.
Please cite our paper if you use this code in your own work.
- python 2.7
- theano
- networkx
- scipy
- Lasagne
python Test.py [dataset][dataset] could be strings: "citeseer", "cora", and "pubmed".
The DGCN model is mainly implemented in Model.py. In layers.py, the dense layer, diffusion layer and dropout function are defined. In LossCalculation.py, the loss calculation functions and evaluation metric function (i.e., accuracy) are defined. In utilities.py, the random walk functions and temporal weight function are defined.
In order to run the code on your own dataset, you need to prepare:
- an n by n adjacency matrix (n is the number of nodes),
- an n by k feature matrix (k is the number of features per node), and
- an n by c binary label matrix (c is the number of classes).
Please refer to our paper and the files DataPreparation.py and utilities.py for detailed data pre-processing information.
For testing, the "citeseer", "cora", and "pubmed" datasets are available in the directory data. Due to the file size limitation, for the "nell_full" dataset, you could find at http://www.cs.cmu.edu/~zhiliny/data/nell_data.tar.gz or our pre-processed version.
Since we used the exactly same datasets for testing, for detailed information about these four datasets, you may refer to the Planetoid repository
Our work is inspired by the following papers:
- Thomas N. Kipf, Max Welling, Semi-Supervised Classification with Graph Convolutional Networks.
- David I Shuman, Sunil K. Narang, Pascal Frossard, Antonio Ortega, Pierre Vandergheynst, The Emerging Field of Signal Processing on Graphs: Extending High-Dimensional Data Analysis to Networks and Other Irregular Domains.
The testing datasets were provided by:
- Zhilin Yang, William W. Cohen, Ruslan Salakhutdinov, Revisiting Semi-Supervised Learning with Graph Embeddings.