A Working Hypothesis on the Semantics/Accuracy Synergy

The main objective of this work is to assess the impact of the initial conditions regarding interpretability and accuracy on the optimizer’s performance and present some guidelines in order to assist the designer. Automatic infusion of anesthesia is a task where both interpretability and accuracy of the controller are highly desirable features. Under this pretext, a description on how these goals can be measured is made and a set of evolutionary optimization schemes is set up. The results show that independently of the optimization process used, the introduction of a semantically valid individual in the initial population is a factor of success in the evolutionary optimization process.