This talk will address the lack of well-labelled representative data associated with the training and verification of Machine Learning.