This study looks at the relationships between different methods of classifier combination and different measures of diversity. We considered ten combination methods and ten measur...
Abstract. Recent findings in the domain of combining classifiers provide a surprising revision of the usefulness of diversity for modelling combined performance. Although there is ...
— We present a methodology to analyze Multiple Classifiers Systems (MCS) performance, using the diversity concept. The goal is to define an alternative approach to the convention...
- An ideal ensemble is composed of base classifiers that perform well and that have minimal overlap in their errors. Eliminating classifiers from an ensemble based on a criterion t...
The disorientation problem has been known to exist on hypertext and hypermedia systems. Various approaches have also been recommended and implemented in order to eliminate or redu...