Research conducted under AFE 87 – “Machine Learning” considered the impact of incorporating machine learning technologies in certifiable aerospace systems. Machine learning and other emerging technologies are fundamentally different than traditional hardware software system and these differences can create difficulties applying existing design assurance guidance, processes, and tools used for certification.
The project was intended to address the following fundamental certification questions:
- What are the performance-based objectives an applicant should satisfy to demonstrate the system with the machine learning meets its intended function in all foreseeable operating conditions, recognizing the foreseeable operating conditions may need to be bounded?
- What are the methods for determining a training data set is a) correct, and b) complete?
- When is machine learning ‘retraining” needed and how is the extent of the retraining determined? How much retraining is required, for example, if changes are made to the sensors, neural net structure, or neural net activation functions, etc.?
- Can the simplex architecture monitors work for all machine learning applications to include the very complex? What are the conditions when a simplex architecture cannot work?
The project produced a report that project members agreed to make available to the public to stimulate further research into new methods supporting certification of systems incorporating machine learning technologies. AVSI is developing a follow-on project to explore specific recommendations outlined in the report. For more information about this project, please contact AVSI.
