pykale 0.1.0b2

Knowledge-aware machine learning from multiple sources in Python

PyKale

---

Getting Started | Documentation | Contributing | Discussions

PyKale is a PyTorch library for multimodal learning and transfer learning on graphs, images, and videos. By adopting a unified pipeline-based API design, PyKale enforces standardization and minimalism. PyKale aims to facilitate interdisciplinary, knowledge-aware machine learning research for graphs, images, and videos in computer vision, graph analysis, and medical imaging applications. It focuses on leveraging knowledge from multiple sources for accurate and interpretable prediction. PyKale's other key machine learning areas of interests include dimensionality reduction and deep learning. See the Trello board for an API overview.

Pipeline-based core API (generic and reusable)

• loaddata loads data from disk or online resources as in input
• prepdata preprocesses data to fit machine learning modules below (transforms)
• embed embeds data in a new space to learn a new representation (feature extraction/selection)
• predict predicts a desired output
• evaluate evaluates the performance using some metrics
• interpret interprets the features and outputs via post-prediction analysis mainly via visualisation
• pipeline specifies a machine learning workflow by combining several other modules

Example usage

• examples demonstrate real applications on specific datasets.

Installation

You should install PyTorch matching your hardware first. To work on graphs, install PyTorch Geometric first follow its official instructions.

Install PyKale using pip for the stable version:

pip install pykale  # for the core kale API only
pip install pykale[extras]  # for Examples/Tutorials (including core API)

Install from source for the latest version and/or development:

git clone https://github.com/pykale/pykale
cd pykale
pip install .  # for core API only
pip install .[extras]  # with extras for examples/tutorials
pip install -e .[dev]  # editable install for developers including all dependencies

To run the unit tests:

pytest

More comprehensive test cases are not yet available.

Examples, Tutorials, and Discussions

See our numerous examples (and tutorials) on how to perform prediction tasks in PyKale.

Ask and answer questions over on PyKale's GitHub Discussions tab.

Contributing

We appreciate all contributions. You can contribute in three ways:

• Star and fork PyKale to follow its latest developments, share it with your networks, and ask questions about it.
• Use PyKale in your project and let us know any bugs (& fixes) and feature requests/suggestions via creating an issue.
• Contribute via branch, fork, and pull for minor fixes and new features, functions, or examples to become one of the contributors.

See contributing guidelines for more details. You can also reach us via email if needed. The participation in this open source project is subject to Code of Conduct.

The Team

PyKale is primarily maintained by a group of researchers at the University of Sheffield: Haiping Lu, Raivo Koot, Xianyuan Liu, Shuo Zhou, and Peizhen Bai.

We would like to thank our other contributors including (but not limited to) Cameron Mcwilliam, Robert Turner, David Jones, and Will Furnass.

Citation

@Misc{pykale2021,
  author =   {Haiping Lu and Raivo Koot and Xianyuan Liu and Shuo Zhou and Peizhen Bai},
  title =    {{PyKale}: Knowledge-aware machine learning from multiple sources in Python},
  howpublished = {\url{https://github.com/pykale/pykale}},
  year = {2021}
}

Acknowledgements

The development of PyKale is partially supported by the following project(s).

• Wellcome Trust Innovator Awards: Digital Technologies Ref 215799/Z/19/Z "Developing a Machine Learning Tool to Improve Prognostic and Treatment Response Assessment on Cardiac MRI Data".