Probability-Enhanced Predictions in the Anticipatory Classifier System

摘要: The Anticipatory Classifier System (ACS) recently showed many capabilities new to the Learning field. Due its enhanced rule structure with an effect part, it forms internal environmental representation, learns latently besides common reward learning, and can use cognitive processes. This paper introduces a probability-enhancement in predictions of ACS which enables system handle different kinds non-determinism environment. Experiments two mazes will show that is now able action-noise irrelevant random attributes perceptions. Furthermore, applications introduced GA reveal general independence mechanism as well ability substantially decrease population size.