This project is a follow on to AFE 87 and will address recommendations raised in the AFE 87 Final Report. The objective is to start to develop practical approaches to assuring the safety of ML-based systems that can be used by system developers and by the regulators charged with certifying the aircraft containing these systems. The proposed scope is to
- Select attributes of ML systems identified in AFE 87 that do not align with current design assurance tools, processes and technologies for deeper investigation.
- Identify and/or develop potential approaches to assuring the safety of ML-based systems
- Select a use case and develop a set of models that will be used to exercise and refine these approaches
- Document the results
- Liaise with other organizations and researchers working in this area
This project is still in development. If you are interested in participating or for more information, contact AVSI.