Book Image

Hands-On Graph Neural Networks Using Python

By : Maxime Labonne
Book Image

Hands-On Graph Neural Networks Using Python

By: Maxime Labonne

Overview of this book

Graph neural networks are a highly effective tool for analyzing data that can be represented as a graph, such as networks, chemical compounds, or transportation networks. The past few years have seen an explosion in the use of graph neural networks, with their application ranging from natural language processing and computer vision to recommendation systems and drug discovery. Hands-On Graph Neural Networks Using Python begins with the fundamentals of graph theory and shows you how to create graph datasets from tabular data. As you advance, you’ll explore major graph neural network architectures and learn essential concepts such as graph convolution, self-attention, link prediction, and heterogeneous graphs. Finally, the book proposes applications to solve real-life problems, enabling you to build a professional portfolio. The code is readily available online and can be easily adapted to other datasets and apps. By the end of this book, you’ll have learned to create graph datasets, implement graph neural networks using Python and PyTorch Geometric, and apply them to solve real-world problems, along with building and training graph neural network models for node and graph classification, link prediction, and much more.

Related Content you might be interested in

Current Title:

Small book image
Hands-On Graph Neural Networks Using Python
Maxime Labonne
Apr, 2023 | 354 pages
Small book image
Graph Machine Learning
Claudio Stamile | Aldo Marzullo | Enrico Deusebio
Jun, 2021 | 338 pages
Small book image
Graph Data Science with Neo4j
Estelle Scifo
Jan, 2023 | 288 pages
Small book image
10 Machine Learning Blueprints You Should Know for Cybersecurity
Rajvardhan Oak
May, 2023 | 330 pages
Table of Contents (25 chapters)
Preface
Preface
Who this book is for
What this book covers
To get the most out of this book
Download the example code files
Download the color images
Conventions used
Get in touch
Share your thoughts
Download a free PDF copy of this book
1
Part 1: Introduction to Graph Learning
Part 1: Introduction to Graph Learning
Free Chapter
2
Chapter 1: Getting Started with Graph Learning
Chapter 1: Getting Started with Graph Learning
Why graphs?
Why graph learning?
Why graph neural networks?
Summary
Further reading
3
Chapter 2: Graph Theory for Graph Neural Networks
Chapter 2: Graph Theory for Graph Neural Networks
Technical requirements
Introducing graph properties
Discovering graph concepts
Exploring graph algorithms
Summary
4
Chapter 3: Creating Node Representations with DeepWalk
Chapter 3: Creating Node Representations with DeepWalk
Technical requirements
Introducing Word2Vec
DeepWalk and random walks
Implementing DeepWalk
Summary
Further reading
5
Part 2: Fundamentals
Part 2: Fundamentals
6
Chapter 4: Improving Embeddings with Biased Random Walks in Node2Vec
Chapter 4: Improving Embeddings with Biased Random Walks in Node2Vec
Technical requirements
Introducing Node2Vec
Implementing Node2Vec
Building a movie RecSys
Summary
Further reading
7
Chapter 5: Including Node Features with Vanilla Neural Networks
Chapter 5: Including Node Features with Vanilla Neural Networks
Technical requirements
Introducing graph datasets
Classifying nodes with vanilla neural networks
Classifying nodes with vanilla graph neural networks
Summary
Further reading
8
Chapter 6: Introducing Graph Convolutional Networks
Chapter 6: Introducing Graph Convolutional Networks
Technical requirements
Designing the graph convolutional layer
Comparing graph convolutional and graph linear layers
Predicting web traffic with node regression
Summary
Further reading
9
Chapter 7: Graph Attention Networks
Chapter 7: Graph Attention Networks
Technical requirements
Introducing the graph attention layer
Implementing the graph attention layer in NumPy
Implementing a GAT in PyTorch Geometric
Summary
10
Part 3: Advanced Techniques
Part 3: Advanced Techniques
11
Chapter 8: Scaling Up Graph Neural Networks with GraphSAGE
Chapter 8: Scaling Up Graph Neural Networks with GraphSAGE
Technical requirements
Introducing GraphSAGE
Classifying nodes on PubMed
Inductive learning on protein-protein interactions
Summary
Further reading
12
Chapter 9: Defining Expressiveness for Graph Classification
Chapter 9: Defining Expressiveness for Graph Classification
Technical requirements
Defining expressiveness
Introducing the GIN
Classifying graphs using GIN
Summary
Further reading
13
Chapter 10: Predicting Links with Graph Neural Networks
Chapter 10: Predicting Links with Graph Neural Networks
Technical requirements
Predicting links with traditional methods
Predicting links with node embeddings
Predicting links with SEAL
Summary
Further reading
14
Chapter 11: Generating Graphs Using Graph Neural Networks
Chapter 11: Generating Graphs Using Graph Neural Networks
Technical requirements
Generating graphs with traditional techniques
Generating graphs with graph neural networks
Generating molecules with MolGAN
Summary
Further reading
15
Chapter 12: Learning from Heterogeneous Graphs
Chapter 12: Learning from Heterogeneous Graphs
Technical requirements
The message passing neural network framework
Introducing heterogeneous graphs
Transforming homogeneous GNNs to heterogeneous GNNs
Implementing a hierarchical self-attention network
Summary
Further reading
16
Chapter 13: Temporal Graph Neural Networks
Chapter 13: Temporal Graph Neural Networks
Technical requirements
Introducing dynamic graphs
Forecasting web traffic
Predicting cases of COVID-19
Summary
Further reading
17
Chapter 14: Explaining Graph Neural Networks
Chapter 14: Explaining Graph Neural Networks
Technical requirements
Introducing explanation techniques
Explaining GNNs with GNNExplainer
Explaining GNNs with Captum
Summary
Further reading
18
Part 4: Applications
Part 4: Applications
19
Chapter 15: Forecasting Traffic Using A3T-GCN
Chapter 15: Forecasting Traffic Using A3T-GCN
Technical requirements
Exploring the PeMS-M dataset
Processing the dataset
Implementing the A3T-GCN architecture
Summary
Further reading
20
Chapter 16: Detecting Anomalies Using Heterogeneous GNNs
Chapter 16: Detecting Anomalies Using Heterogeneous GNNs
Technical requirements
Exploring the CIDDS-001 dataset
Preprocessing the CIDDS-001 dataset
Implementing a heterogeneous GNN
Summary
Further reading
21
Chapter 17: Building a Recommender System Using LightGCN
Chapter 17: Building a Recommender System Using LightGCN
Technical requirements
Exploring the Book-Crossing dataset
Preprocessing the Book-Crossing dataset
Implementing the LightGCN architecture
Summary
Further reading
22
Chapter 18: Unlocking the Potential of Graph Neural Networks for Real-World Applications
Chapter 18: Unlocking the Potential of Graph Neural Networks for Real-World Applications
23
Index
Index
Why subscribe?
24
Other Books You May Enjoy
Other Books You May Enjoy
Packt is searching for authors like you
Share your thoughts
Download a free PDF copy of this book

To get the most out of this book

You should have a basic understanding of graph theory and machine learning concepts, such as supervised and unsupervised learning, training, and the evaluation of models to maximize your learning experience. Familiarity with deep learning frameworks, such as PyTorch, will also be useful, although not essential, as the book will provide a comprehensive introduction to the mathematical concepts and their implementation.

Software covered in the book

Operating system requirements

Python 3.8.15

Windows, macOS, or Linux

PyTorch 1.13.1

Windows, macOS, or Linux

PyTorch Geometric 2.2.0

Windows, macOS, or Linux

To install Python 3.8.15, you can download the latest version from the official Python website: https://www.python.org/downloads/. We strongly recommend using a virtual environment, such as venv or conda.

Optionally, if you want to use a Graphics Processing Unit (GPU) from NVIDIA to accelerate training and inference, you will need to install CUDA and cuDNN:

CUDA is a parallel computing platform and API developed by NVIDIA for general computing on GPUs. To install CUDA, you can follow the instructions on the NVIDIA website: https://developer.nvidia.com/cuda-downloads.

cuDNN is a library developed by NVIDIA, which provides highly optimized GPU implementations of primitives for deep learning algorithms. To install cuDNN, you need to create an account on the NVIDIA website and download the library from the cuDNN download page: https://developer.nvidia.com/cudnn.

You can check out the list of CUDA-enabled GPU products on the NVIDIA website: https://developer.nvidia.com/cuda-gpus.

To install PyTorch 1.13.1, you can follow the instructions on the official PyTorch website: https://pytorch.org/. You can choose the installation method that is most appropriate for your system (including CUDA and cuDNN).

To install PyTorch Geometric 2.2.0, you can follow the instructions in the GitHub repository: https://pytorch-geometric.readthedocs.io/en/2.2.0/notes/installation.html. You will need to have PyTorch installed on your system first.

Chapter 11 requires TensorFlow 2.4. To install it, you can follow the instructions on the official TensorFlow website: https://www.tensorflow.org/install. You can choose the installation method that is most appropriate for your system and the version of TensorFlow you want to use.

Chapter 14 requires an older version of PyTorch Geometric (version 2.0.4). It is recommended to create a specific virtual environment for this chapter.

Chapter 15, Chapter 16, and Chapter 17 require a high GPU memory usage. You can lower it by decreasing the size of the training set in the code.

Other Python libraries are required in some or most chapters. You can install them using pip install <name==version>, or using another installer depending on your configuration (such as conda). Here is the complete list of required packages with the corresponding versions:

  • pandas==1.5.2
  • gensim==4.3.0
  • networkx==2.8.8
  • matplotlib==3.6.3
  • node2vec==0.4.6
  • seaborn==0.12.2
  • scikit-learn==1.2.0
  • deepchem==2.7.1
  • torch-geometric-temporal==0.54.0
  • captum==0.6.0

The complete list of requirements is available on GitHub at https://github.com/PacktPublishing/Hands-On-Graph-Neural-Networks-Using-Python. Alternatively, you can directly import notebooks in Google Colab at https://colab.research.google.com.

If you are using the digital version of this book, we advise you to type the code yourself or access the code from the book’s GitHub repository (a link is available in the next section). Doing so will help you avoid any potential errors related to the copying and pasting of code.

Previous Section
Next Section