By training with virtual opponents in simulators, trainee fighter pilots can build the experience necessary for air combat operations, at a fraction of the cost of training with real aircraft. In practice however, the variety of the opponents is not as wide as it can be. This is largely due to a lack of behaviour models. In his PhD research (2013-2020), Armon Toubman (NLR) investigated the extent to which it is possible to generate the required behaviour models, by means of machine learning. In this webinar he discusses the intersection of training simulations, behaviour models, and machine learning, and shares the results of his research.