|
|
Chris Thornton
The paper presents the results of an empirical study in which supervised learning algorithms were used to train an animat to perform a difficult navigation task. The results of the study are explained in terms of a theoretical distinction between two learning-complexity classes.