A reduced probabilistic neural network for the classification of large databases

The probabilistic neural network (PNN) is a special type of radial basis neural network used mainly for classification problems. Due to the size of the network after training, this type of network is usually used for problems with a small-sized training dataset. In this paper, a new training algorithm is presented for use with large training databases. Application to the handwritten digit database shows that the reduced PNN performs better than the standard PNN for all of the studied cases with a big gain in size and processing speed. This new type of neural network can be used easily for problems with large training databases like biometrics and data mining applications. An extension of the network is possible for new training samples and/or classes without retraining.

A reduced probabilistic neural network for the classification of large databases

The probabilistic neural network (PNN) is a special type of radial basis neural network used mainly for classification problems. Due to the size of the network after training, this type of network is usually used for problems with a small-sized training dataset. In this paper, a new training algorithm is presented for use with large training databases. Application to the handwritten digit database shows that the reduced PNN performs better than the standard PNN for all of the studied cases with a big gain in size and processing speed. This new type of neural network can be used easily for problems with large training databases like biometrics and data mining applications. An extension of the network is possible for new training samples and/or classes without retraining.

___

  • samples to represent each class. The rest of the training set is considered as noisy redundant samples. The classification rate for the training set is 100%, which is similar to the standard PNN.
  • The drawback with the RPNN is that it needs more time to train since training is an evaluation of the contribution of each neuron in the hidden layer. Training a standard PNN is simply copying the training samples in the hidden layer, and this is done in one pass. This is why the standard PNN is only used for small databases. 7. Conclusion
  • In this paper, a new training algorithm is proposed for PNNs. The proposed algorithm reduces the size of a PNN while increasing accuracy and processing speed.
  • This algorithm can be used for all cases where a standard PNN suffers from the size of the training dataset and it is preferable for real-world large databases, where the gain in processing speed becomes very important.
  • T.P. Tran, T.T.S. Nguyen, P. Tsai, X. Kong, “BSPNN: boosted subspace probabilistic neural network for email security”, Artificial Intelligence Review, Vol. 35, pp. 369–382, 2011.
  • T.P. Tran, L. Cao, D. Tran, C.D. Nguyen, “Novel intrusion detection using probabilistic neural network and adaptive boosting”, International Journal of Computer Science and Information Security, Vol. 6, pp. 083–091, 2009.
  • M.R. Nikoo, R. Kerachian, S. Malakpour-Estalaki, S.N. Bashi-Azghadi, M.M. Azimi-Ghadikolaee, “A probabilistic water quality index for river water quality assessment: a case study”, Environmental Monitoring and Assessment,
  • Vol. 181, pp. 465–478, 2010.
  • M.S. Bascil, H. Oztekin, “A study on hepatitis disease diagnosis using probabilistic neural network”, Journal of Medical Systems, Vol. 36, pp. 1603–1606, 2010.
  • M. Hariharan, M.P. Paulraj, S. Yaacob, “Time-domain features and probabilistic neural network for the detection of vocal fold pathology”, Malaysian Journal of Computer Science, Vol. 23, pp. 60–67, 2010.
  • F. Budak, E.D. ¨Ubeyli, “Detection of resistivity for antibiotics by probabilistic neural networks”, Journal of Medical Systems, Vol. 35, pp. 87–91, 2009.
  • M.I. Faraj, J. Bigun, “Synergy of lip-motion and acoustic features in biometric speech and speaker recognition”, IEEE Transactions on Computers, Vol. 56, pp. 1169–1175, 2007.
  • S. Meshoul, M. Batouche, “A novel approach for online signature verification using fisher based probabilistic neural network”, IEEE Symposium on Computers and Communications, pp. 314–319, 2010.
  • K.T. Blackwell, T.P. Vogl, H.P. Dettmar, M.A. Brown, G.S. Barbour, D.L. Alkon, “Identification of faces obscured by noise: comparison of an artificial neural network with human observers”, Journal of Experimental & Theoretical
  • Artificial Intelligence, Vol. 9, pp. 491–508, 1997.
  • L.F. Araghi, H. Khaloozade, M.R. Arvan, “Ship identification using probabilistic neural networks (PNN)”, Inter- national MultiConference of Engineers and Computer Scientists, Vol. 2, 2009.
  • M.W. Kim, M. Arozullah, “Generalized probabilistic neural network based classifiers”, International Joint Confer- ence on Neural Networks, 2002, Vol. 3, pp. 648–653, 1992.
  • P. Burrascano, “Learning vector quantization for the probabilistic neural network”, IEEE Transactions on Neural Networks, Vol. 2, pp. 458–461, 1991.
  • V. Georgiou, P. Alevizos, M. Vrahatis, “Fuzzy evolutionary probabilistic neural networks”, Artificial Neural Net- works in Pattern Recognition, Vol. 5064, pp. 113–124, 2008.
  • I. De Falco, A. Della Cioppa, E. Tarantino, “Facing classification problems with particle swarm optimization”, Applied Soft Computing, Vol. 7, pp. 652–658, 2007.
  • I. Galleske, J. Castellanos, “A rotated kernel probabilistic neural network (RKPNN) for multi-class classification”, Proceedings of the Artificial and Natural Neural Networks 7th International Conference on Computational Methods in Neural Modeling, Vol. 1, pp. 152–157, 2003.
  • T.P. Hong, S.S. Tseng, “A probabilistic perceptron learning algorithm”, Journal of Experimental & Theoretical Artificial Intelligence, Vol. 4, pp. 265–279, 1992.
  • R.O. Duda, Pattern Classification 2nd Edition with Computer Manual 2nd Edition Set. New York, Wiley- Interscience, 2004.
  • D.K. Kim, D.H. Kim, S.K. Chang, “Modified probabilistic neural network considering heterogeneous probabilistic density functions in the design of breakwater”, Korean Society of Civil Engineers Journal of Civil Engineering, Vol. 11, pp. 65–71, 2007.
  • D.F. Specht, “Probabilistic neural networks”, Neural Networks, Vol. 3, pp. 109–118, 1990.
  • N. Neggaz, M. Besnassi, A. Benyettou, “Application of improved AAM and probabilistic neural network to facial expression recognition”, Journal of Applied Sciences, Vol. 10, pp. 1572–1579, 2010.
  • J. Li, K. Ouazzane, H. Kazemian, Y. Jing, R. Boyd, “A neural network based solution for automatic typing errors correction”, Neural Computer & Applications, Vol. 20, pp. 889–896, 2011.