Using Machine Learning to extract Models of human Control Skills

Experimental results are presented of experiments on using Machine Learning algorithms to extract a model of a human control skill. The control task to be modeled is that of a pilot flying a F-16 flightsimulation on alevelled left turn. Machine Learning algorithms used include C4.5 as a decision tree representative and FOIL and Indlog as Inductive Logic Programming representatives. Improvements on the methodology of previous experiments are presented. An extension to decision trees is proposed to solve some of the shortcomings. ILP systems are being used to improve the extended decision tree results