Develop the machine learnt model using the development data obtained in the previous stage such that the allocated ML safety requirements are satisfied.
Use internal test data to assess the extent to which the machine learnt model is able to meet the ML safety requirements when presented with data not used for development.
Create an assurance argument, based on the evidence generated by meeting the first two objectives, which provides a clear justification that the ML model meets the ML safety requirements. This should explicitly explain the tradeoffs, assumptions and uncertainties concerning both the ML model and the process by which it is developed and validated.
As shown in the AMLAS model learning assurance process diagram above, this stage consists of three activities. The artefacts generated from this stage are used to instantiate the ML model assurance argument pattern as part of Activity 12.
Additional guidance on this stage can be found at .