/gnn-receptive-fields

Repository for my 2021 research internship at Dascim Polytechnique

Primary LanguageJupyter Notebook

GNN: Linearly growing Receptive Fields

Repository for my 2021 research internship at Dascim Polytechnique

Benchmark

Models

  • GAT: Graph Attention Neural Networks (torch_geometric.nn.models.GAT)
  • GCN: Graph Convolutional Neural Networks (torch_geometric.nn.models.GCN)
  • JK-Net: Jumping Knowledge Network (torch_geometric.nn.models.JumpingKnowledge)
  • AdaGCN: Adaboost with GCN
  • MADReg & AdaEdge

Datasets

  • Short-range graphs
    • Cora
    • CiteSeer
    • Amazon (Product Classification)
    • Reddit (Subreddit prediction)
  • Long-range graphs
    • QM9
    • ENZYMES
    • NCI1

Colab Snipets

  • Using .py modules hosted in github:
!wget <github_path_to_python_module>
  • Installing packages once:
import os, sys
import os.path as osp
from google.colab import drive
drive.mount('/content/mnt')
nb_path = '/content/notebooks'
os.symlink('/content/mnt/My Drive/Colab Notebooks', nb_path)
sys.path.insert(0, nb_path)

!pip install --target=$nb_path __package_name__