Skip to content

[PAKDD 2021] Graph InfoClust: Leveraging cluster-level node information for unsupervised graph representation learning

Notifications You must be signed in to change notification settings

cmavro/Graph-InfoClust-GIC

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

32 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Graph-InfoClust-GIC [PAKDD 2021]

PAKDD'21 version Graph InfoClust: Maximizing Coarse-Grain Mutual Information in Graphs

Preprint version Graph InfoClust: Leveraging cluster-level node information for unsupervised graph representation learning

PWC PWC PWC

Overview

GIC’s framework. (a) A fake input is created based on the real one. (b) Embeddings are computed for both inputs with a GNN-encoder. (c) The graph and cluster summaries are computed. (d) The goal is to discriminate between real and fake samples based on the computed summaries.

gic-dgl

GIC (node classification task) implemented in Deep Graph Library (DGL) , which should be installed.

python train.py --dataset=[DATASET]

GIC

GIC (link prediction, clustering, and visualization tasks) based on Deep Graph Infomax (DGI) original implementation.

python execute_link.py

Cite

@misc{mavromatis2020graph,
    title={Graph InfoClust: Leveraging cluster-level node information for unsupervised graph representation learning},
    author={Costas Mavromatis and George Karypis},
    year={2020},
    eprint={2009.06946},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

or

@inproceedings{Mavromatis2021GraphIM,
  title={Graph InfoClust: Maximizing Coarse-Grain Mutual Information in Graphs},
  author={Costas Mavromatis and G. Karypis},
  booktitle={PAKDD},
  year={2021}
}

About

[PAKDD 2021] Graph InfoClust: Leveraging cluster-level node information for unsupervised graph representation learning

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published