Next: Learning Algorithm: Questions Up: Building the Software Agent Previous: Behavior Mechanism Including Feedback Contents
If the above described feedback mechanism is working, then it is possible to use the accumulating feedback values to modify the existing classifiers. The overall goal is to generate a set of classifiers which map the available perceptions in an optimal way into a set of actions which enable the system to reach its goals sufficiently well. This part of the theory is not yet realized. This will be done during the next weeks.