Selection of a suitable classifier fusion scheme in the design of multiple classifier systems (MCSs) is a tedious task. To meet this we propose a neuro-fuzzy fusion (NFF) method for fusing the responses of a set of fuzzy classifiers. In the proposed method the output of the considered classifiers are fed to a neural network which performs the fusion task. Five labeled data sets, of which two are from remote sensing images, have been used for the performance comparison of various MCSs. Experimental study revealed the improved classification capability of the proposed NFF based MCS yielding consistently better results for all data sets. © 2006 IEEE.
Neuro-Fuzzy Fusion: A New Approach to Multiple Classifier System
Bruzzone, Lorenzo
2006-01-01
Abstract
Selection of a suitable classifier fusion scheme in the design of multiple classifier systems (MCSs) is a tedious task. To meet this we propose a neuro-fuzzy fusion (NFF) method for fusing the responses of a set of fuzzy classifiers. In the proposed method the output of the considered classifiers are fed to a neural network which performs the fusion task. Five labeled data sets, of which two are from remote sensing images, have been used for the performance comparison of various MCSs. Experimental study revealed the improved classification capability of the proposed NFF based MCS yielding consistently better results for all data sets. © 2006 IEEE.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione
