The week's pick
Random Articles
Reseach Article
Pruned Modified Fuzzy Hyperline Segment Neural Network and Its Application to Pattern Classification
| International Journal of Computer Applications |
| Foundation of Computer Science (FCS), NY, USA |
| Volume 93 - Number 12 |
| Year of Publication: 2014 |
| Authors: S. B. Bagal, U. V. Kulkarni |
10.5120/16271-6008
|
S. B. Bagal, U. V. Kulkarni . Pruned Modified Fuzzy Hyperline Segment Neural Network and Its Application to Pattern Classification. International Journal of Computer Applications. 93, 12 ( May 2014), 43-50. DOI=10.5120/16271-6008
Abstract
The Fuzzy hyperline segment neural network (FHLSNN) is supervised classifier that forms n-dimensional hyperline segments (HLS) defined by two end points with a corresponding membership function for learning and testing. In this paper, the Pruned fuzzy hyperline segment neural network (PFHLSNN) and Pruned modified fuzzy hyperline segment neural network (PMFHLSNN) are proposed. The pruning method is based on a confidence factor calculated for each hyperline segment in the prediction phase after learning. The new definition of confidence factor is proposed. In PFHLSNN, the hyperline segments with low confidence factor are pruned using user defined threshold to reduce the network complexity. In order to improve the classification performance of PFHLSNN, the modification is proposed in its testing phase and the network is referred as PMFHLSNN. In this modification, the Euclidean distance is computed between the applied input pattern and the centroid of the patterns falling on the hyperline segment to decide the class of pattern. Finally, the HLS with smallest distance is selected as winner and the pattern is so classified that it belongs to the class associated with that HLS. The performance of PFHLSNN and PMFHLSNN is evaluated using benchmark problems and real world handwritten character recognition data set. The results are analyzed, discussed and compared with the FHLSNN. Thus, the proposed approach improved the classification accuracy without affecting the incremental learning of FHLSNN and reduces the network complexity by pruning the hyperline segments of low confidence factor.
References
- J. J. Buckley and Y. Hayashi, "Fuzzy neural network: a survey," Journal of Fuzzy Sets and Systems, vol. 66, Issue 1, pp. 1-13, Aug. 1994.
- P. K. Simpson, "Fuzzy min–max neural networks—Part 1: Classification," Journal of IEEE Trans. Neural Networks. , vol. 3, no. 5, pp. 776–786, Sep. 1992.
- P. K. Simpson, "Fuzzy min–max neural networks—Part 2: Clustering," IEEE Trans. Fuzzy systems, vol. 1, no. 1, pp. 32–45, Feb. 1993.
- B. Gabrys and A. Bargiela, "General Fuzzy Min-Max Neural Network for Clustering and Classification," IEEE Trans. neural networks, vol. 11, pp. 769-783, May 2000.
- U. V. Kulkarni, T. R. Sontakke, and G. D. Randale, 2001. Fuzzy hyperline segment neural for rotation invariant handwritten character recognition. In Proceeding of the Int. joint conf. on neural networks: IJCCNN'01 held in Washington DC, USA, pp. 2918–2923.
- U. V. Kulkarni, T. R. Sontakke and A. B. Kulkarni, "Fuzzy Hyperline Segment Clustering Neural Network," In IEEE Electronics letters vol. 37, no. 05, pp. 301-303, March 2001.
- P. M. Patil and T. R. Sontakke, 2006. "Rotation, scale and translation invariant handwritten Devanagari numeral character recognition using general fuzzy Neural Network" Journal of Pattern Recognition, vol. 40, pp. 2110–2117, 2007.
- Sietsma J. and Dow R. J. F. , 1988. Neural Net Pruning-why and how. In Proceeding of Int. Joint Conf. on Neural Networks held in San Diego, C. A. USA. , Vol. 1, pp. 325-333.
- R. Reed, "Pruning algorithms-A survey," Journal of IEEE Trans. Neural Networks, vol. 4, pp. 740-747, 1993.
- R. Setiono, "A penalty function approach for pruning Feedforward neural networks", Journal of Neural Computation, vol. 9, no. 1, pp. 185-204, 1997.
- G. Carpenter and A. Tan, "Rule extraction: From neural architecture to symbolic representation," Journal of Connection Sci. , vol. 7, no. 1, pp. 3–27, 1995.
- S. M. Kamruzzaman and Ahmed Ryadh Hasan, 2005. Pattern classification using simplified neural networks with pruning algorithm. In Proceeding of Int. Joint Conf. on Information and Communication Technology in Management, Multimedia University, Malaysia.
- Anas Quteishat and Chee Peng Lim, 2008. Application of the fuzzy Min-Max neural networks to medical diagnosis. In Proceeding of 12th International Conference on Knowledge-Based and Intelligent Information & Engineering Systems , KES2008, Zagreb, Croatia, Part III, pp 548-555.
- Anas M. Quteishat and Chee Peng Lim, "A Modified fuzzy min-max neural network and its application to fault classification," Journal of Soft computing in industrial applications, ASC, vol. 39, pp. 179-188, 2007.
- Johnson, R. A. , and D. W. Wichern, 1998. Applied multivariate Statistical Analysis. New Jersey: Prentice Hall. pp. 226-235.
- P. M. Murphy and D. W. Aha, 1995. UCI Repository of Machine Learning Databases, (Machine-Readable Data Repository). Irvine, CA: Dept. Inf. Computer Sci. , Univ. California.
- Perantonis S. J. and P. J. G. Lisboa, "Translation, rotation and scale invariant pattern recognition by high-order neural networks and moment classifiers," Journal of IEEE Trans. Neural networks, Vol. 3, No. 2, 1992, pp. 241-251.
- Udea K. and Y. Nakamura, 1984. Automatic verification of sealimpression pattern. In Proceeding of 9th Int. Conf. on pattern recognition. Vo1. 2, pp. 1019-1021.
- Hung-Pin Chiu and Din-Chang Tseng, "Invariant handwritten Chinese character recognition using fuzzy min-max neural networks," Pattern Recog. Letters, Vol. 18, 1997, pp. 481-491.
- A. Hoang, 1997, Supervised classifier performance on the UCI data set, in Department of Computer Science, vol. M. Sc. , University of Adelaide, Australia.
- Anas Q. , C. P. Lim, and K. S. Tan, "A modi?ed fuzzy min–max neural network with a genetic-algorithm-based rule extractor for pattern classi?cation". IEEE Trans. Systems, Man, and Cybernetics—part a: systems and humans, vol. 40, no. 3, May 2010.
Index Terms
Keywords