Previous |  Up |  Next


handwritten digits; combined classifiers
Classifiers can be combined to reduce classification errors. We did experiments on a data set consisting of different sets of features of handwritten digits. Different types of classifiers were trained on these feature sets. The performances of these classifiers and combination rules were tested. The best results were acquired with the mean, median and product combination rules. The product was best for combining linear classifiers, the median for $k$-NN classifiers. Training a classifier on all features did not result in less errors.
[1] Breukelen M. van, Duin R. P. W., Tax D. M. J., Hartog J. E. den: Combining classifiers for the recognition of handwritten digits. In: Proceedings of the 1st IAPR TC1 Workshop on Statistical Techniques in Pattern Recognition, Prague 1997, pp. 13–18
[2] Duin R. P. W.: PRTOOLS, A Matlab Toolbox for Pattern Recognition. 1995
[3] al M. D. Garris et: NIST Form–Based Handprint Recognition System. Internal Report National Institute of Standards and Technology NISTIR 5469, 1994
[4] Hartog J. E. den, Kate T. K. ten, Gerbrands J. J.: Knowledge–based interpretation of utility maps. In: Computer Vision Graphics and Image Processing: Image Understanding, 1996, pp. 105–117
[5] Khotanzad A., Hong Y. H.: Rotation invariant pattern recognition using Zernike moments, In: Int. Conf. on Pattern Recognition, Rome 1998, pp. 326–328
[6] Kittler J., Hatef M., Duin R. P. W.: Combining classifiers. In: Proceedings of ICPR’96, pp. 897–901
[7] Tumer K., Ghosh J.: Theoretical Foundations of Linear and Order Statistics Combiners for Neural Pattern Classifiers. TR-95-02-98, The Computer and Vision Research Center, The University of Texas at Austin, 1995
Partner of
EuDML logo