egc 0.1.2

Benchmark of Graph Clustering.

EAGLEGraphClustering

A benchmark of community detection algorithms from EAGLE-Lab, ZJU

• Usage
• Datasets
• Implemented baseline methods
  • Disjoint
    • Unsupervised Graph Neural Networks + Kmeans
    • End-to-End Community Detection
  • Overlapping
• Installation
• Requirements

Usage

Run any supported model by :

$ python train.py ${OPTIONAL global args} ${POSITIONAL args (model)} ${optional model args}

• OPTIONAL global args which should be used before ${model}

  global args	description
  -h, --help	show this help message and exit
  --dataset	Dataset used in the experiment
  --gpu	ID of gpu used by cuda
  --dir	Path to store the dataset
  --seed	Random seed. Defaults to 4096.
  --not_set_seed	Force Not to Use Random Seed.

• POSITIONAL args, i.e., all models supported

  positional args (models supported)	description
  pca_kmeans	run community detection on PCA with kmeans
  SDCN	run community detection on SDCN
  DANMF	run community detection on DANMF
  MNMF	run community detection on MNMF
  gae_kmeans	run community detection on GAE with Kmeans
  vgae_kmeans	run community detection on VGAE with Kmeans
  VGAECD	run community detection on VGAECD
  dgi_kmeans	run community detection on DGI with Kmeans
  gmi_kmeans	run community detection on GMI with Kmeans
  SENet_kmeans	run community detection on SENet with Kmeans
  DFCN	run community detection on DFCN
  CommunityGAN	run community detection on CommunityGAN
  ComE	run community detection on ComE

e.g.,

$ python train.py --dataset=Cora gmi_kmeans --l2_coef=0.0

Check all optional args of a model by python train.py ${model} -h.

Datasets

• Cora, Citeseer, Pubmed come from DGL Lib
• BlogCatalog, Flickr come from CoLA github
• ACM come from SDCN github
• All above datasets are converted to undirected graphs.

Dataset	Nodes	Edges	Attributes	Classes
Cora	2708	10556	1433	7
Citeseer	3327	9228	3703	6
Pubmed	19717	88651	500	3
BlogCatalog	5196	343486	8189	6
Flickr	7575	479476	12047	9
ACM	3025	26256	1870	3
CoraFull	19793	126842	8710	70

Implemented baseline methods

Disjoint

Unsupervised Graph Neural Networks + Kmeans

method	Conf/Journal	Original Code	Supproted
VGAE	16nips	TensorFlow	✅
GraphSAGE
DGI	19iclr	Pytorch	✅
GMI	20www	Pytorch	✅
SENet	21Neural Networks		✅

End-to-End Community Detection

method	Conf/Journal	Original Code	Supproted
SDCN	20www	Pytorch	✅
DANMF	18cikm	code	✅
M-NMF	17aaai	Matlab	✅
DFCN	21aaai	Pytorch	✅
VGAECD	18icdm	--	✅
ComE	17cikm	code	✅
DAEGC	19ijcai	code	✅

Overlapping

method	Conf/Journal	Original Code	Supproted
CommunityGAN	19www	TensorFlow	✅

Installation

# NOTE: python>=3.8 is needed
# Install cuda if necessary. Check Cuda version first.
$ sudo bash .ci/cuda.sh
# Leave out `bash .ci/install-dev.sh &&` if no dev env is needed.
$ bash .ci/install-dev.sh && bash .ci/install.sh
# run `source .env/bin/activate` to activate the virtual env

Requirements

DGL installation guide: https://www.dgl.ai/pages/start.html

torch installation guide: https://pytorch.org/get-started/locally/

• CUDA 11

python >= 3.8
cudatoolkit >= 11.1.1
torch >= 1.12.0
dgl >= 1.1.0
ogb >= 1.3.3
pylint >= 2.11.1
pre-commit >= 2.15.0
yapf >= 0.31.0
tqdm >= 4.62.3
texttable >= 1.6.4
Cython >= 0.25