A new fuzzy approach for multi-source decision fusion

Nowadays, we are facing the rapidly growing amount of data being produced in many organizations, social networks and internet. These data are generated in disparate locations and their aggregation into one location is exceedingly time and space consuming. Traditional statistical methods are not sufficient for processing of this massive multi-source data. In this paper, we propose a new fuzzy-based decision fusion approach for classification problems of this kind. The necessity of fuzzy information arises in distributed classification because imprecision, uncertainty and ambiguity can be found at all information sources, from the data itself to the results of the classifiers. In the proposed approach, multiple classifiers are constructed based on different information sources which have different degrees of reliability. Then a fuzzy rule based system is designed for approximating distribution of reliabilities of sources over the input space. The decision fusion of multiple classifiers takes place using the estimated degrees of sources' reliabilities. Comparison results are made between both centralized classification and two other distributed classification methods. One is averaging and the other is discounting each classifier's decision based on its accuracy. Results show the high accuracy of the proposed method in making decisions in distributed environments, without the overhead of aggregating the entire data in one location.

[1]  Tong Lee,et al.  Probabilistic and Evidential Approaches for Multisource Data Analysis , 1987, IEEE Transactions on Geoscience and Remote Sensing.

[2]  Ludmila I. Kuncheva,et al.  Switching between selection and fusion in combining classifiers: an experiment , 2002, IEEE Trans. Syst. Man Cybern. Part B.

[3]  Hakan Cevikalp,et al.  Local Classifier Weighting by Quadratic Programming , 2008, IEEE Transactions on Neural Networks.

[4]  Josef Kittler,et al.  Multiple Classifier Fusion in Probabilistic Neural Networks , 2002, Pattern Analysis & Applications.

[5]  Kevin W. Bowyer,et al.  Combination of Multiple Classifiers Using Local Accuracy Estimates , 1997, IEEE Trans. Pattern Anal. Mach. Intell..

[6]  Catherine Blake,et al.  UCI Repository of machine learning databases , 1998 .

[7]  Johannes R. Sveinsson,et al.  A classifier ensemble based on fusion of support vector machines for classifying hyperspectral data , 2010 .

[8]  B. Jeon,et al.  Decision fusion with reliabilities in multisource data classification , 1992, [Proceedings] 1992 IEEE International Conference on Systems, Man, and Cybernetics.

[9]  Ronald R. Yager,et al.  Set Measure Directed Multi-Source Information Fusion , 2011, IEEE Transactions on Fuzzy Systems.

[10]  Sargur N. Srihari,et al.  Decision Combination in Multiple Classifier Systems , 1994, IEEE Trans. Pattern Anal. Mach. Intell..

[11]  Jiri Matas,et al.  On Combining Classifiers , 1998, IEEE Trans. Pattern Anal. Mach. Intell..

[12]  Jon Atli Benediktsson,et al.  Fusion of Support Vector Machines for Classification of Multisensor Data , 2007, IEEE Transactions on Geoscience and Remote Sensing.

[13]  D. J. Newman,et al.  UCI Repository of Machine Learning Database , 1998 .

[14]  Chih-Jen Lin,et al.  A Practical Guide to Support Vector Classication , 2008 .

[15]  P. Swain,et al.  Neural Network Approaches Versus Statistical Methods In Classification Of Multisource Remote Sensing Data , 1990 .

[16]  Piyaphol Phoungphol,et al.  A Classification Framework for Imbalanced Data , 2013 .

[17]  Jon Atli Benediktsson,et al.  Classification of multisource and hyperspectral data based on decision fusion , 1999, IEEE Trans. Geosci. Remote. Sens..

[18]  Chengqi Zhang,et al.  Combined Mining: Discovering Informative Knowledge in Complex Data , 2011, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics).