How Strength and Accuracy Differ
We begin this Chapter with a short discussion of the difficulties of working with and understanding complex systems such as classifier systems, in which we suggest that many models or paradigms of a complex system may need to be considered before we arrive at the most suitable ones. Following this we consider a number of models of classifier systems. We review the rationale behind Holland’s classifier systems and XCS, as presented by their authors. We do not engage in detailed discussions of their ideas, but rather attempt to state in simple terms why they expect these systems to learn. The main point of this part of the Chapter will be that strength and accuracy-based systems differ fundamentally in how they solve problems.
KeywordsBoolean Function Reward Function Sequential Task Accuracy Criterion Learning Classifier System
Unable to display preview. Download preview PDF.