Multiple Binary Classifier Fusion (MBCF) in Application of Satimage Database

R. Ebrahimpour, S.Z. Moussavi, and S.R. Ehteram (Iran)


Satimage , Neural Networks; combining classifier; Decomposition


Multiple Binary Classifier Fusion (MBCF) may generate more accurate classification than each of constituent classifiers .In this paper fusion is based on max rule .We present here a simple rule for adapting the class combiner to the application. To have an analysis on satimage Data base, we present a Binary classifier for system composed of several separate networks. The inputs to Multilayer Perceptrons are feature vectors obtained by Principal components analysis (PCA). the designed method with the proposed of sat image while providing the faster convergence in the training phase requires the hidden layer with a fewer neurons and less sensitivity to the training and testing patterns The efficiency of proposed method is demonstrated on sat image database and comparison with other algorithms yields excellent accuracy rate in Pattern recognition courses Percent rate of 97.3% for sat image database was obtained using the mentioned devised algorithm.

Important Links:

Go Back