Constructing andtraining feed -forwardneural networks for pattern classi$cation

A new approach of constructing andtraining neural networks for pattern classi$cation is proposed. Data clusters are generated andtrainedsequentially basedon d istinct local subsets of the training d ata. Obtainedclusters are then usedto construct a feed -forwardnetwork, which is further trainedusing stand ardalgorithms operating on the global training set. The network obtained using this approach e6ectively inherits the knowledge from the local training procedure before improving on its generalization ability through the subsequent global training. Various experiments demonstrate the superiority of this approach over competing methods. ? 2002 Pattern Recognition Society. Published by Elsevier Science Ltd. All rights reserved.

[1]  D. F. Specht,et al.  Enhancements to probabilistic neural networks , 1992, [Proceedings 1992] IJCNN International Joint Conference on Neural Networks.

[2]  Tom Downs,et al.  CARVE-a constructive algorithm for real-valued examples , 1998, IEEE Trans. Neural Networks.

[3]  Rajeev Kumar,et al.  Multiobjective genetic algorithm partitioning for hierarchical learning of high-dimensional pattern spaces: a learning-follows-decomposition strategy , 1998, IEEE Trans. Neural Networks.

[4]  J. MacQueen.  Some methods for classification and analysis of multivariate observations , 1967 .

[5]  Scott E. Fahlman,et al.  An empirical study of learning speed in back-propagation networks , 1988 .

[6]  Andrew R. Webb,et al.  Shape-adaptive radial basis functions , 1998, IEEE Trans. Neural Networks.

[7]  Dietmar Heinke,et al.  Comparing neural networks: a benchmark on growing neural gas, growing cell structures, and fuzzy ARTMAP , 1998, IEEE Trans. Neural Networks.

[8]  Witold Pedrycz,et al.  Conditional fuzzy clustering in the design of radial basis function neural networks , 1998, IEEE Trans. Neural Networks.

[9]  Noel E. Sharkey,et al.  An Analysis of Catastrophic Interference , 1995, Connect. Sci..

[10]  John Moody,et al.  Fast Learning in Networks of Locally-Tuned Processing Units , 1989, Neural Computation.

[11]  Sven Behnke,et al.  Competitive neural trees for pattern classification , 1998, IEEE Trans. Neural Networks.

[12]  Mohamad T. Musavi,et al.  On the training of radial basis function classifiers , 1992, Neural Networks.

[13]  Bernard Widrow,et al.  Improving the learning speed of 2-layer neural networks by choosing initial values of the adaptive weights , 1990, 1990 IJCNN International Joint Conference on Neural Networks.

[14]  Hans G. C. Tråvén,et al.  A neural network approach to statistical pattern classification by 'semiparametric' estimation of probability density functions , 1991, IEEE Trans. Neural Networks.

[15]  Martin A. Riedmiller,et al.  A direct adaptive method for faster backpropagation learning: the RPROP algorithm , 1993, IEEE International Conference on Neural Networks.

[16]  C. L. Scofield,et al.  Into silicon: real time learning in a high density RBF neural network , 1991, IJCNN-91-Seattle International Joint Conference on Neural Networks.

[17]  Pietro Burrascano,et al.  Learning vector quantization for the probabilistic neural network , 1991, IEEE Trans. Neural Networks.

[18]  K. Lang,et al.  Learning to tell two spirals apart , 1988 .

[19]  Thierry Denoeux,et al.  Initializing back propagation networks with prototypes , 1993, Neural Networks.

[20]  Christian Lebiere,et al.  The Cascade-Correlation Learning Architecture , 1989, NIPS.

[21]  Teuvo Kohonen,et al.  The self-organizing map , 1990 .

[22]  D. L. Reilly,et al.  A neural model for category learning , 1982, Biological Cybernetics.