GNN / GCN Guide

d9249·2022년 3월 25일


목록 보기

GNN Lectures

  • CS224W by Jure Leskovec, [Link]

GNN frameworks

GNN architectures

Review papers

  • The Graph Neural Network Model, [Link]
  • Geometric deep learning: going beyond Euclidean data, [Link], [YouTube]
  • Relational inductive biases, deep learning, and graph networks, [Link]

Important papers

  • Convolutional Networks on Graphs for Learning Molecular Fingerprints, [Link], [GitHub]
  • Gated Graph Sequence Neural Networks, [Link]
  • Semi-Supervised Classification with Graph Convolutional Networks, [Link], [GitHub]
  • Neural Message Passing for Quantum Chemistry, [Link], [GitHub]
  • Graph Attention Networks, [Link], [GitHub]
  • How Powerful are Graph Neural Networks?, [Link]

Graph generative models

  • Learning Deep Generative Models of Graphs, [Link], [Github]
  • MolGAN: An implicit generative model for small molecular graphs, [Link], [GitHub]
  • Junction Tree Variational Autoencoder for Molecular Graph Generation, [Link], [GitHub]
  • Constrained Graph Variational Autoencoders for Molecule Design, [Link], [GitHub]
  • Graph Convolutional Policy Network for Goal-Directed Molecular Graph Generation, [Link], [GitHub]
  • DEFactor: Differentiable Edge Factorization-based Probabilistic Graph Generation, [Link],
  • Learning Multimodal Graph-to-Graph Translation for Molecular Optimization, [Link], [GitHub]
  • Efficient Graph Generation with Graph Recurrent Attention Networks, [Link], [GitHub]
  • Efficient learning of non-autoregressive graph variational autoencoders for molecular graph generation, [Link], [GitHub]

Pooling methods

Node/Graph classification

  • Hyperbolic Graph Convolutional Neural Networks, [Link]
  • Hyperbolic Graph Neural Networks, [Link], [GitHub]

Link prediction and recommender system

Molecular applications

  • SchNet: A continuous-filter convolutional neural network for modeling quantum interactions, [Link]
  • MoleculeNet: a benchmark for molecular machine learning, [Link]
  • Bayesian graph convolutional neural networks for semi-supervised classification, [Link]

Physics modeling

  • Interaction Networks for Learning about Objects, Relations and Physics, [Link]
  • Neural Relational Inference for Interacting Systems, [Link]
  • Learning Symbolic Physics with Graph Networks, [Link]
  • Structural Recurrent Neural Network (SRNN) for Group Activity Analysis, [Link]
  • H-OGN


AI researcher가 되고싶은 석사 연구생입니다.

0개의 댓글