Distance-based Disagreement Classifiers Combination

We present a methodology to analyze multiple classifiers systems (MCS) performance, using the diversity concept. The goal is to define an alternative approach to the conventional recognition rate criterion, which usually requires an exhaustive combination search. This approach defines a distance-based disagreement (DbD) measure using an Euclidean distance computed between confusion matrices and a soft-correlation rule to indicate the most likely candidates to the best classifiers ensemble. As case study, we apply this strategy to two different handwritten recognition systems. Experimental results indicate that the method proposed can be used as a low-cost alternative to conventional approaches.

[1]  Ludmila I. Kuncheva,et al.  Measures of Diversity in Classifier Ensembles and Their Relationship with the Ensemble Accuracy , 2003, Machine Learning.

[2]  Terry Windeatt,et al.  Diversity measures for multiple classifier system analysis and design , 2004, Inf. Fusion.

[3]  Yves Lecourtier,et al.  Controlling the diversity in classifier ensembles through a measure of agreement , 2005, Pattern Recognit..

[4]  Cinthia Obladen de Almendra Freitas,et al.  Handwritten recognition with multiple classifiers for restricted lexicon , 2004, Proceedings. 17th Brazilian Symposium on Computer Graphics and Image Processing.

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

[6]  Ching Y. Suen,et al.  A class-modular feedforward neural network for handwriting recognition , 2002, Pattern Recognit..

[7]  Jim R. Parker,et al.  Algorithms for image processing and computer vision , 1996 .

[8]  Biing-Hwang Juang,et al.  Fundamentals of speech recognition , 1993, Prentice Hall signal processing series.

[9]  Luiz Eduardo Soares de Oliveira,et al.  Handwritten Character Recognition Using Nonsymmetrical Perceptual Zoning , 2007, Int. J. Pattern Recognit. Artif. Intell..

[10]  Robert P. W. Duin,et al.  The characterization of classification problems by classifier disagreements , 2004, Proceedings of the 17th International Conference on Pattern Recognition, 2004. ICPR 2004..

[11]  Ludmila I. Kuncheva,et al.  Moderate diversity for better cluster ensembles , 2006, Inf. Fusion.

[12]  Luiz Eduardo Soares de Oliveira,et al.  Zoning and metaclasses for character recognition , 2007, SAC '07.

[13]  Cinthia Obladen de Almendra Freitas,et al.  Methodology for the design of NN-based month-word recognizers written on Brazilian bank checks , 2007, Image Vis. Comput..

[14]  Ching Y. Suen,et al.  Analysis and recognition of alphanumeric handprints by parts , 1992, Proceedings., 11th IAPR International Conference on Pattern Recognition. Vol.II. Conference B: Pattern Recognition Methodology and Systems.

[15]  Cinthia Obladen de Almendra Freitas,et al.  Study Of Perceptual Similarity Between Different Lexicons , 2004, Int. J. Pattern Recognit. Artif. Intell..

[16]  Ching Y. Suen,et al.  A regional decomposition method for recognizing handprinted characters , 1995, IEEE Trans. Syst. Man Cybern..