GAE VGAE Save

These are the graph embedding methods that I reproduce.

1. There are four tasks used to evaluate the effect of embeddings, i.e., node clustering, node classification, link_prediction, and graph Visualization. Algorithms used in the tasks:

• Clustering：k-means;
• Classification: SVM;
• Link_Prediction;
• Visualization: t-SNE;

2. Requirement: Python 3.7, Pytorch: 1.5 and other pakeages which is illustrated in the code.

3. There are two types of datasets: graph datasets(cora, citeseer, pubmed) and no-graph datasets(att,imm,umist,orl). If you want to use other datasets, you just need to put your dataset in the "Dataset" folder. The adjacency matrix for the no-graph datasets is calculated by the KNN (k=9).

4. There are three classical graph datasets (cite, citeseer, pubmed) for the graph convolution network(GNNs). Besides, some figures datasets such ATT, COIL and IMM are also included in this file. Moreover, I don't only provide the datasets but also offer the methods to preprocess those datasets. These datasets can be used for many tasks such as clustering, classification and Link prediction, etc. The datasets and methods are useful and friendly to the beginners.

5. The reproduced algorithm for paper is as follows:

• GAE_VGAE: T. N. Kipf, M. Welling, "Variational graph auto-encoders," arXiv preprint arXiv:1611.07308, 2016.
• LGAE_LVGAE: G. Salha, R. Hennequin, and M. Vazirgiannis, "Keep it simple: Graph autoencoders without graph convolutional networks," arXiv preprint arXiv:1910.00942, 2019.
• SDNE: D. Wang, P. Cui, and W. Zhu, "Structural deep network embedding," in Proc. IEEE Conf. Proceedings of the 22nd ACM SIGKDD international conference on Knowledge discovery and data mining, 2016, pp. 1225-1234.

6. If you think my code is useful to you, please give me a little star, thank you