The Best Practices

The aim of the Best Practices is to be easily accessible to anyone working on or interested in machine learning. They are designed for a large audience who come from a variety of backgrounds and organisations, and who have different needs. Therefore, their implementation can be flexible when the context is accounted for.

About the Best Practices

The Best Practices are designed to be adaptable to different organisation sizes, needs, risks, resources, and expected societal impact. They also aim to be complete, and therefore can be a bit long at times. Luckily, depending on the situation, their implementation can be flexible as long as the following 3 pillars are kept in mind:

  • The Technical and Organisational aspects of the Best Practices are equally important and are aligned;

  • The Best Practices cover most risks that are associated with machine learning. Depending on the organisation's context, some of these risks will be more relevant than others; and

  • The entire Product Lifecycle requires full attention when attending to the Best Practices and their implementation.

What can the Best Practices be used for?

  • A way to get started with implementing ethical and responsible machine learning in products;

  • A common language between data scientists, engineers, managers and governance & compliance professionals;

  • A repository for your responsible machine learning questions;

  • A point of reference for your machine learning audits, policies, governance and regulations.

The Best Practices

The Best Practices are split into two parts to help tackle the different elements of ethical and responsible machine learning more easily.

The Organisation Best Practice

FBPML_OrganisationBP_V1.0.pdf

Scoped for the entire organisation. It details how to effectively support machine learning product teams within an organisation.

The Technical Best Practice

FBPML_TechnicalBP_V1.0.pdf

Scoped for a single product (including machine learning models). It will help data science teams to develop and maintain machine learning products ethically and responsibly.

The Wiki portal contains the same Best Practices as the releases on this website, but the Wiki versions also have the latest community contributions added to them. Anyone can submit, view and/or discuss contributions on the Wiki at all times: it is the community hub. Contributions will be officially incorporated periodically into the releases.


The Wiki portal also serves as a platform to find and contribute supporting materials to the Best Practices. This might take place in the form of examples or shared experiences.


Read more about the FBPML Wiki

What is next for the Best Practices?

You can expect us to keep improving the Best Practices. This includes not only their content, but also their readability and operationalisation. Work is currently underway for additional supporting material, such as situation-specific implementation advice. Upgrades are also planned for the Wiki portal and community platform in the near future.

Note that the Best Practices have been licensed under the Creative Commons Attribution license. This means they are freely available for commercial and/or private adaptation and/or use, subject to attributing The Foundation of Best Practices of Machine Learning. Should you have any more questions in this regard, please feel free to contact us.