We are looking forward to seeing and participating in an active discussion by the AI community. Shaping the future regulation policy taking into account the industry response, is an important initiative of EASA.
If you want to get a short outline of the essential points of the document, check out this article on Avionics International titled EASA Awaits Industry Feedback on Initial Regulatory Guidance for Machine Learning Algorithms in Aircraft Systems. (We are pleased to see that our participation in the EASA effort referenced in the document didn't go unnoticed).
Also, the EASA AI Task Force members (Guillaume Soudain, François Triboulet and Jean Marc Cluzeau) gave a detailed overview of the document and outlined what EASA will require from the ML application developers during the future certification process on the 4th EUROCONTROL's FlyAI webinar; we recommend watching it:
The key presentations are available for download on the webinar page.
For Daedalean, it was specifically exciting to watch the presentations describing the practical use cases implementing the Learning Assurance concept. In particular, we were pleased to learn from Benjamin Cramet how Eurocontrol NM LAB applies the W-Shape to their project on optimisation of the climb and descent. Read our blog post with an explanation of the W-shaped framework for learning assurance in simple words.