/LER

Primary LanguagePython

LER

This repository is the implementation of our project of SJTU-CS222 Algorithm Design & Analysis:

LER：Spectral Data Augmentation for Graph Neural Network Wenyi Xue, Sijia Li, Luyuan Jin

learnable LPA

Use the first part of GCN-LPA, an end-to-end model that unifies Graph Convolutional Neural Networks (GCN) and Label Propagation Algorithm (LPA) for adaptive semi-supervised node classification, proposed by Hongwei Wang and Jure Leskovec in the paper .

Files in the folder

  • data/
    • citeseer/
    • cora/
    • pubmed/
    • ms_academic_cs.npz (Coauthor-CS)
    • ms_academic_phy.npz (Coauthor-Phy)
  • src/: implementation of LER.

Sparsification Algorithms of unweighted/weighted Graph

Graph Sparsification by Effective Resistances, Daniel A. Spielman, Nikhil Srivastava, 2009. A Unifying Framework for Spectrum-Preserving Graph Sparsification and Coarsening, Gecia Bravo-Hermsdorff, Lee M. Gunderson, 2020.

Running the code

$ python main.py

Note: The default dataset is Citeseer. Hyper-parameter settings for other datasets are provided in main.py.

Required packages

The code has been tested running under Python 3.6.5, with the following packages installed (along with their dependencies):

  • tensorflow == 1.12.0
  • networkx == 2.1
  • numpy == 1.14.3
  • scipy == 1.1.0
  • sklearn == 0.19.1
  • matplotlib == 2.2.2