Our Story¶
The XML Book – and, more generally, the XMLX organisation – was concocted by Kacper Sokol to deliver a comprehensive outlook on the most popular explainability methods in Machine Learning. Our vision is to build a growing collection of resources covering both theory and practice in a range of different difficulty levels, including but not limited to:
- high-level overviews & introductory examples;
- mathematical foundations;
- algorithmic implementations;
- practical advice & real-life caveats; and
- success & failure case studies.
The book is intended as a slowly maturing repository of relevant resources – as opposed to one-shot content delivery – so expect the materials to evolve over time as the field progresses and matures.
Breadcrumbs¶
The XML Book builds upon our prior work and experience in eXplainable Machine Learning, such as:
- FAT Forensics – a Python package for inspecting Fairness, Accountability and Transparency of predictive pipelines;
- What and How of Machine Learning Transparency – a hands-on Machine Learning eXplainability tutorial organised at ECML-PKDD 2020;
- Explainability Fact Sheets – a comprehensive overview of social and technical properties relevant to eXplainable Machine Learning; and
- Kacper’s research and, more broadly, work done by the Intelligent Systems Laboratory at the University of Bristol.
Contributors¶
The book development is led by Kacper Sokol who together with his colleagues from the University of Bristol – Alexander Hepburn and Torty Sivill – prepared and published its first version. A list of contributors based on GitHub statistics is displayed below. With the exception of occasional funding, the book is developed and maintained by volunteers passionate about explainability of data-driven predictive models. We are always looking for contributors; if you are interested, please have a look at our development page or reach out.
Funding¶
Below you can find an overview of our supporters. The initial funding for the book was provided by The Alan Turing Institute under their Online Training Call (more details about the award).
