The Pattern Classification Based on Fuzzy Min-max Neural Network with New Algorithm
A new fuzzy min-max neural network (FMNN) based on based on new algorithm is proposed for pattern classification. A new membership function of hyperbox is defined in which the characteristic are considered. The FMNN with new learning algorithm don’t use contraction process of fuzzy min-max neural network described by Simpson.The new algorithm only need expansion and no additional neurons have been added to the neural network to deal with the overlapped area. FMNN with new algorithm has strong robustness and high accuracy in classification for considering the characteristic of data core and noise. The performance of FMNN with new algorithm is checked by some benchmark data sets and compared with some traditional methods. All the results indicate that FMNN with new algorithm is effective. abstract environment.
Keywordsfuzzy min-max neural network patter classification robustness learning algorithm
Unable to display preview. Download preview PDF.
- 10.Alpern, B., Carter, L.: The hyperbox. In: Proc. IEEE Conf. Visualization, pp. 133–139 (October 1991)Google Scholar
- 11.Rizzi, A., Panella, M., Mascioli, F.M.F.: A recursive algorithm for fuzzy min-max networks. In: Proc. IEEE/INNS/ENNS. International Joint Conference Neural Networks (IJCNN 2000), vol. 6, pp. 541–546 (July 2000)Google Scholar
- 17.Blake, C., Keogh, E., Merz, C.J.: UCI Repository of Machine Learning Database University of California, Irvine (1998), http://www.ics.uci.edu/~mlearn/MLRepositroy.html