ABSTRACT
The term multiple models or ensemble of classifiers is used to identify a set of classifiers for which individual decisions are in some way combined (typically by voting) to classify new examples (Dietterich, 1997). The main idea behind any multiple model system is based on the observation that different learning algorithms explore different representation languages, search spaces, and evaluation functions of the hypothesis. How can we explore these differences? Is it possible to design a set of classifiers that working together can obtain a better performance than each individual classifier? Multiple models are also used in the context of dynamic streams, where the target concept may change over time.
