Demos

• Basics
  • Loading data into StellarGraph from NetworkX
  • Loading data into StellarGraph from Pandas
  • Loading and saving data between StellarGraph and Neo4j
• Calibration
  • Link prediction example and calibration: GraphSAGE on the PubMed-Diabetes citation dataset
  • Node Attribute Inference (multi-class) using GraphSAGE and the Pubmed-Diabetes citation network with calibration
• Community Detection
  • Comparison of clustering of node embeddings with a traditional community detection method
• Connector
  • Neo4j
    • Stellargraph example: directed CORA citation network with directed GraphSAGE and Neo4J.
    • Notebook demo on loading Cora dataset into Neo4J Database
    • Undirected GraphSAGE on the CORA citation network using StellarGraph and Neo4j
• Embeddings
  • GCN Deep Graph Infomax on CORA
  • Comparing Different Models
  • Graph representation learning through Unsupervised GraphSAGE
  • GraphWave aglorithm on a barbell graph
  • Node Representation Learning with attri2vec on Citeseer
  • Introduction
  • Load the dataset
  • The Metapath2Vec algorithm
  • Representation Learning using Word2Vec
  • Visualise Node Embeddings
  • Downstream task
  • Introduction
  • Corpus generation using random walks
  • Loading in the data
  • Creating the model
  • Training
  • Visualizing Embeddings
• Ensembles
  • StellarGraph Ensemble for link prediction
  • Stellargraph Ensembles for node attribute inference
• Graph Classification
  • Supervised graph classification example
• Interpretability
  • Gat
    • Interpreting Nodes and Edges by Saliency Maps in GAT
    • Node and link importance via saliency maps
  • Gcn
    • Intepretability on Hateful Twitter Datasets
    • Loading the CORA network
    • Splitting the data
    • Converting to numeric arrays
    • Creating the GCN model in Keras
    • Training the model
    • Loading the CORA network
    • Splitting the data
    • Converting to numeric arrays
    • Creating the GCN model in Keras
    • Training the model
• Link Prediction
  • Attri2vec
    • Inductive Node Representation Learning through attri2vec
  • Gcn
    • Link prediction example: GCN on the Cora citation dataset
  • Graphsage
    • Link prediction example: GraphSAGE on the Cora citation dataset
  • Hinsage
    • Stellargraph example: Heterogeneous GraphSAGE on the Movielens recommendation dataset
  • Knowledge Graphs
    • ComplEx on WN18 and FB15K
    • DistMult on WN18 and FB15K
  • Random Walks
    • Link prediction with Node2Vec on Cora
    • Load the dataset
    • Construct splits of the input data
    • Node2Vec
    • Train and evaluate the link prediction model
    • Continuous-Time Dynamic Network Embeddings
• Node Classification
  • Attri2vec
    • Node Representation Learning with attri2vec on Citeseer
  • Cluster Gcn
    • Cluster-GCN for node classification
  • Gat
    • Stellargraph example: Graph Attention Network (GAT) on the CORA citation dataset
  • Gcn
    • Graph Convolutional Network (GCN) on the CORA citation dataset
  • Graphsage
    • Stellargraph example: Fully directed GraphSAGE on a directed CORA citation network
    • Stellargraph example: GraphSAGE on the CORA citation network
    • Inductive representation learning using GraphSAGE on the Pubmed-Diabetes citation network
  • Node2vec
    • Introduction
    • Corpus generation using random walks
    • Node2Vec with weighted random walks
  • Ppnp
    • Stellargraph example: Personalised Propagation of Neural Predictions (PPNP) and Approximate PPNP (APPNP) on the CORA citation dataset
  • Rgcn
    • Stellargraph example: Relational Graph Convolutional Network (RGCN) on the AIFB relational dataset
  • Sgc
    • Stellargraph example: Simplified Graph Convolutions (SGC) on the CORA citation dataset
• Use Cases
  • Detection of Twitter users who use hateful lexicon using graph machine learning with Stellargraph