This web page offers an introduction to the tools, techniques and overall issues relating to combining classifiers into ensembles. Those looking for more information may wish to consult the references included in the bibliography. The book of Ludmila Kuncheva (2004) offers an excellent survey of the field and served as the primary reference for this web page. A more succinct but much less comprehensive overview can be found in the fifteenth chapter of Alpaydin’s book (2004). The International Workshops on Multiple Classifier Systems, held since 2000, has served as a key venue for the presentation of valuable research in this domain.
This web site is organized into a number of sections. Although an informed reader should feel free to jump to sections of particular interest, those new to the study of ensemble classification may benefit from reading the sections sequentially, as they do build upon one another to a certain extent. The sections are organized as follows:
Section 1: A general introduction to the classifier
ensembles.
Section 2: Reasons for using classifier ensembles
as opposed to single classifiers.
Section 3: An overview of general issues to consider
when designing classifier ensembles.
Section 4: An overview of the techniques
that can be use to combine classifiers into ensembles. These are discussed in
more detail in Sections 5 and 6.
Section 5: Combining classifiers using
fusion.
Section 6: Combining classifiers using
selection.
Section 7: Some additional methodologies
that have proven themselves to be effective, including bagging and boosting.
Section 8: The importance of classifier diversity.
Section 9: Some concluding remarks.
Section 10: References.
Section 11: An applet demonstrating the role of diversity
in ensemble classification.
Next: Introduction
Last modified: April 20, 2005.
-top of page-