Integrating multiple classifiers by finding their areas of expertise

Using multiple learned classifers for increasing learning accuracy has attracted much recent interest. A central problem is how to integrate several classifers to produce a single fnal classifcation. We have explored an approach which relies on analyzing the particular area of expertise of each classifer. In this paper, we describe a method which uses inductive techniques on a set of training examples to determine the areas of a particular classifer's expertise. By applying this method to each of a set of con icting classifers, we can determine the most reliable classifer for each new example and use that classifer to classify the example. In this way, we achieve signifcantly better classifcation accuracy than even the best individual expert or an induced combination rule.