Bulent Bolat
Electronics and Telecommunication Engineering Department
Yildiz Technical University, Besiktas, Istanbul 34349, Turkey
Tulay Yildirim
Electronics and Telecommunication Engineering Department
Yildiz Technical University, Besiktas, Istanbul 34349, Turkey
ABSTRACT
Through this paper, some performance increasing methods for probabilistic neural network (PNN) are presented. These methods are tested with the glass benchmark database which has an irregular class distribution. Selection of a good training dataset is one of the most important issue. Therefore, a new data selection procedure was proposed. A data replication method is applied to the rare events of the dataset. After reaching the best accuracy, a principal component analysis (PCA) is used to reduce the computational complexity of PNN. Better classification accuracy than the reference work using Bayesian EM model was achieved by PNN using these methods.
PDF References Citation
How to cite this article
Bulent Bolat and Tulay Yildirim, 2003. Performance Increasing Methods for Probabilistic Neural Networks. Information Technology Journal, 2: 250-255.
DOI: 10.3923/itj.2003.250.255
URL: https://scialert.net/abstract/?doi=itj.2003.250.255
DOI: 10.3923/itj.2003.250.255
URL: https://scialert.net/abstract/?doi=itj.2003.250.255
REFERENCES
- Avci, M. and T. Yildirim, 2002. Classification of E. coli bacteria by artificial neural networks. Proc. IEEE Int. Symp. Intell. Syst., 3: 13-16.
CrossRefDirect Link - Ganchev, T., N. Fakotakis and G. Kokkinakis, 2002. Speaker verification system based on probabilistic neural networks. Proceedings of the NIST Speaker Recognition Evaluation Workshop, May 19-22, Virginia, USA., pp: 1-2.
Direct Link - Goh, T.C., 2002. Probabilistic neural network for evaluating seismic liquefaction potential. Can. Geotech. J., 39: 219-232.
Direct Link - Romero, R., D. Touretzky and R.H. Thibadeau, 1997. Optical Chinese character recognition using probabilistic neural networks. Pattern Recognition, 30: 1279-1292.
CrossRefDirect Link - Ruda, H. and M. Snorasson, 1995. Adaptive preprocessing for on-line learning with adaptive resonance theory (ART) networks. Proceedings of the IEEE Workshop on Neural Networks for Signal Processing, Aug. 31-Sept. 2, Cambridge, MA, USA., pp: 513-520.
Direct Link - Specht, D.F., 1990. Probabilistic Neural Networks. Vol. 3, Elsevier Science Ltd., Oxford, UK., pp: 109-118.
Direct Link