Feature Membership Functions in Voronoi-Based Zoning
Recently, the problem of zoning design has been considered as an optimization problem and the optimal zoning is found as the one which minimizes the value of the cost function associated to the classification. For the purpose, well-suited zoning representation techniques based on Voronoi Diagrams have been proposed and effective real-coded genetic algorithms have been used for optimization.
In this paper, starts from the consideration that whatever zoning method is considered, the role of feature membership function is crucial, since it determines the influence of a feature to each zone of the zoning method. Thus, in the paper the role of feature membership functions in Voronoi-based zoning methods is investigated. For the purpose, abstract-level, ranked-level and measurement-level membership functions are considered and their effectiveness is estimated under different Voronoi-based zoning methods.
The experimental tests, carried out in the field of hand-written numeral recognition, show that the best results are obtained when specific measurement-level membership functions are used.
KeywordsGenetic Algorithm Membership Function Voronoi Diagram Voronoi Region Optimal Zoning
Unable to display preview. Download preview PDF.
- 4.Dimauro, G., Impedovo, S., Modugno, R., Pirlo, G.: Numeral recognition by weighting local decisions. In: Proc. ICDAR 2003, Edinburgh, UK, pp. 1070–1074 (August 2003)Google Scholar
- 7.Kuncheva, L.I., Jain, L.C.: Designing Classifier Fusion Systems by Genetic Algorithms. IEEE Trans. Evolut. Comput. 4(4) (2000)Google Scholar
- 10.Baeck, T.: Evolutionary Algorithms in Theory and Practice: Evolution Strategies, Evolution Programming, Genetic Algorithms. Oxford Univ. Press, New York (1996)Google Scholar
- 11.Michalewicz, Z.: Genetic Algorithms + Data Structure=Evolution Programs. Springer, Berlin (1996)Google Scholar
- 12.Grefenstette, J.J.: Optimization of control parameters for genetic algorithms. IEEE Trans. SMC 16(1), 122–128 (1986)Google Scholar
- 13.Back, T., Fogel, D., Michalewicz, Z. (eds.): Handbook of Evolutionary Computation. Institute of Physics Publishing Ltd., Bristol and Oxford University Press, New York (1997)Google Scholar
- 14.Hull, J.: A database for handwritten text recognition research. IEEE Trans. PAMI 16(5), 550–554 (1994)Google Scholar
- 16.Naccache, N.J., Shinghal, R.: SPTA: A proposed algorithm for thinning binary patterns. IEEE Trans. SMC 14(3), 409–418 (1994)Google Scholar