Ensemble approaches have been shown to enhance classification by combining the outputs from a set of voting classifiers. Diversity in error patterns among base classifiers promotes ensemble performance. Multi-task learning is an important characteristic for Neural Network classifiers. Introducing a secondary output unit that receives different training signals for base networks in an ensemble can effectively promote diversity and improve ensemble performance. Here a Competitive Learning Neural Network Ensemble is proposed where a secondary output unit predicts the classification performance of the primary output unit in each base network. The networks compete with each other on the basis of classification performance and partition the stimulus space. The secondary units adaptively receive different training signals depending on the competition. As the result, each base network develops ¡°preference¡± over different regions of the stimulus space as indicated by their secondary unit outputs. To form an ensemble decision, all base networks¡¯ primary unit outputs are combined and weighted according to the secondary unit outputs. The effectiveness of the proposed approach is demonstrated with the experiments on one real-world and four artificial classification problems.
| Identifer | oai:union.ndltd.org:PITT/oai:PITTETD:etd-03102010-152349 |
| Date | 12 May 2010 |
| Creators | Ye, Qiang |
| Contributors | Hassan Karimi, Satish Iyengar, Paul Munro, Stephen Hirtle, Bambang Parmanto |
| Publisher | University of Pittsburgh |
| Source Sets | University of Pittsburgh |
| Language | English |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | http://etd.library.pitt.edu/ETD/available/etd-03102010-152349/ |
| Rights | unrestricted, I hereby certify that, if appropriate, I have obtained and attached hereto a written permission statement from the owner(s) of each third party copyrighted matter to be included in my thesis, dissertation, or project report, allowing distribution as specified below. I certify that the version I submitted is the same as that approved by my advisory committee. I hereby grant to University of Pittsburgh or its agents the non-exclusive license to archive and make accessible, under the conditions specified below, my thesis, dissertation, or project report in whole or in part in all forms of media, now or hereafter known. I retain all other ownership rights to the copyright of the thesis, dissertation or project report. I also retain the right to use in future works (such as articles or books) all or part of this thesis, dissertation, or project report. |
Page generated in 0.2746 seconds