Ensembles for Supervised Classification Learning
- March 1997.
- Physical description
- 100 p.
Also available at
At the library
Media & Microtext Center
- Corporate Author
- Stanford University. Computer Science Department CS-TR-97-1587.
- Matan, Ofer
- ["This dissertation studies the use of multiple classifiers (ensembles or committees) in learning tasks. Both theoretical and practical aspects of combining classifiers are studied."]
- ["First we analyze the representational ability of voting ensembles. A voting ensemble may perform either better or worse than each of its individual members. We give tight upper and lower bounds on the classification performance of a voting ensemble as a function of the classification performances of its individual members."]
- ["Boosting is a method of combining multiple \"weak\" classifiers to form a \"strong\" classifier. Several issues concerning boosting are studied in this thesis. We study SBA, a hierarchical boosting algorithm proposed by Schapire, in terms of its representation and its search. We present a rejection boosting algorithm that trades-off exploration and exploitation: It requires fewer pattern labels at the expense of lower boosting ability."]
- ["Ensembles may be useful in gaining information. We study their use to minimize labeling costs of data and to enable improvements on performance over time. For that purpose a model for on-site learning is presented. The system learns by querying \"hard\" patterns while classifying \"easy\" ones."]
- Supplemental links
- Thesis (Ph.D.).
- Publication date
- Bib-Version: CS-TR-v2.0
- [Adminitrivia V1/Prg/19970324].
- Technical Report