Outline (Level 2) 2 Multiple classifier combination ●Reason o Architecture 10/55
Outline (Level 2) 2 Multiple classifier combination Reason Architecture 10 / 55
2.1.Reason o Better classification performance than individual classifiers o More resilience to noise Beside avoiding the selection of the worse classifier under particular hypothesis,fusion of multiple classifiers can improve the performance of the best individual classifiers This is possible if individual classifiers make"different"errors For linear combiners,Turner and Ghosh(1996)showed that averaging outputs of individual classifiers with unbiased and uncorrelated errors can improve the performance of the best individual classifier 11/55
2.1. Reason Better classification performance than individual classifiers More resilience to noise Beside avoiding the selection of the worse classifier under particular hypothesis, fusion of multiple classifiers can improve the performance of the best individual classifiers This is possible if individual classifiers make ”different” errors For linear combiners, Turner and Ghosh (1996) showed that averaging outputs of individual classifiers with unbiased and uncorrelated errors can improve the performance of the best individual classifier 11 / 55
Outline (Level 2) Multiple classifier combination o Reason o Architecture 12/55
Outline (Level 2) 2 Multiple classifier combination Reason Architecture 12 / 55
2.2.Architecture parallel serial hybrid 13/55
2.2. Architecture 13 / 55
Architecture-different feature sets F1 Classifier 1 level-2'features F2 Classifier 2 Different feature sets Combining Classifier F3 Classifier 3 F4 Classifier 4 14/55
Architecture - different feature sets 14 / 55