A fuzzy-input fuzzy-output SVM technique for classification of hyperspectral remote sensing images

Abstract

In this paper we present a novel fuzzy input-fuzzy output support vector machine (F <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> -SVM) classifier, which is able to process fuzzy information given as input to the classification algorithm and to produce fuzzy classification outputs. The main novelties of the proposed F <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> -SVM consist of: i) simultaneous and proper management of both uncertainty and fuzzy information; ii) capability to model one-to-many relations between a pattern and the related information classes both in the learning and in the classification phases; iii) capability to address multiclass problems in a fuzzy framework. Experimental results obtained on a hyperspectral data set confirm the effectiveness of the proposed technique, which provided classification accuracies higher than those exhibited by a fuzzy multilayer perceptron neural network classifier used for comparisons