Welcome to Spektral

Spektral is a Python library designed for graph deep learning, built on the Keras API and TensorFlow 2. Its primary goal is to provide a simple yet flexible framework for creating graph neural networks (GNNs).

With Spektral, you can tackle various tasks such as:

• Classifying users in a social network
• Predicting molecular properties
• Generating new graphs using GANs
• Clustering nodes
• Predicting links

Spektral implements several popular layers for graph deep learning, including:

• Graph Convolutional Networks (GCN)
• Chebyshev convolutions
• GraphSAGE
• ARMA convolutions
• Edge-Conditioned Convolutions (ECC)
• Graph attention networks (GAT)
• Approximated Personalized Propagation of Neural Predictions (APPNP)
• Graph Isomorphism Networks (GIN)
• Diffusional Convolutions

Additionally, Spektral provides various pooling layers, such as:

• MinCut pooling
• DiffPool
• Top-K pooling
• Self-Attention Graph (SAG) pooling
• Global pooling
• Global gated attention pooling
• SortPool

For more information on getting started with Spektral, check out the Getting Started guide and explore the examples for templates.

Installation

Spektral is compatible with Python 3.6 and above, and has been tested on the latest versions of Ubuntu, MacOS, and Windows. Other Linux distributions should also work.

The simplest way to install Spektral is via PyPi:

pip install spektral

To install Spektral from source, run the following commands in your terminal:

git clone https://github.com/danielegrattarola/spektral.git
cd spektral
python setup.py install  # Or 'pip install .'

For installation on Google Colab, use:

! pip install spektral

New Features in Spektral 1.0

The 1.0 release of Spektral marks a significant milestone, introducing numerous features and improvements:

• The new Graph and Dataset containers standardize data handling, simplifying usage.
• The Loader class abstracts graph batch creation, allowing focus on training logic.
• The transforms module offers a variety of common graph operations.
• New GeneralConv and GeneralGNN classes enable building versatile models.
• New datasets: QM7 and ModelNet10/40, along with a wrapper for OGB datasets.
• Major library structure and dependency clean-up.
• New examples and tutorials to assist users.

Usage Examples and API Overview

Spektral provides a rich API for building and training graph neural networks. Here’s a simple example of how to create a GCN model:

from spektral.layers import GCNConv
from spektral.models import GeneralGNN

class MyGNN(GeneralGNN):
    def __init__(self):
        super().__init__()
        self.conv1 = GCNConv(16, activation='relu')
        self.conv2 = GCNConv(2)

    def call(self, inputs):
        x, a = inputs
        x = self.conv1([x, a])
        x = self.conv2([x, a])
        return x

This example demonstrates how to define a simple GNN model using Spektral’s layers. For more detailed examples, refer to the examples section.

Community and Contribution

Spektral is an open-source project, and contributions are highly encouraged. You can contribute by:

• Reporting bugs by opening an issue.
• Requesting features through issues or discussions.
• Submitting pull requests for new features or bug fixes.

For detailed contribution guidelines, visit the contributing guidelines.

License and Legal Considerations

Spektral is licensed under the MIT License, allowing for free use, modification, and distribution. The full license text is available in the repository.

For more information, refer to the GitHub repository.

Conclusion

Spektral is a powerful tool for anyone looking to delve into graph deep learning. With its extensive features and user-friendly API, it simplifies the process of building and training graph neural networks. Whether you’re a researcher or a developer, Spektral provides the tools you need to succeed in your projects.

For more information, visit the Spektral GitHub repository and start your journey into graph neural networks today!

FAQ