Update This repo is deprecated. Please refer to https://github.com/tigergraph/graph-ml-notebooks for more recent docs.
Note: The main branch is under active development and might contain breaking changes since the latest release. For docs most relavant to a specific version of the ML Workbench, please check out the corresponding branch, e.g., the 1.0 branch for ML Workbench v1.0.x.
The repository is broken down into tutorials and demos. The tutorials directory contains basic notebooks to get you up and running with the TigerGraph ML Workbench, including basic data loading and manipulation functions, building GNN models, and then finally deploying your models to cloud providers such as Azure and Google Cloud Platform. The demos directory contains notebooks that apply the basic concepts in the tutorials to real-world data and business problems.
The tutorial directory is broken down to three subdirectories: basics, gnn_pyg, cloud_deployment, and advanced. The basics directory contains notebooks that are designed to get developers familiar with the different utilities that pyTigerGraph and the TigerGraph Machine Learning Workbench provide. Moving from that, the gnn_pyg directory contains notebooks that combine the data loading utilities introduced in the basics directory and uses them to train Graph Neural Networks built using PyTorch Geometric. The cloud_deployment directory contains notebooks that apply the concepts in the tutorials to deploy GNN models to cloud providers such as Azure and Google Cloud Platform. Finally, the advanced directory contains relativaly complex uses of ML workbench such as hyperparameter tuning.
There are currently two demos available: ethereum_fraud and recommendation. etheruem_fraud walks you through training a Graph Neural Network to detect fraudulent Ethereum accounts. recommendation walks you through training a Graph Neural Network to recommend music to users in the LastFM dataset.