The learning dynamics of a Radial Basis Function (RBF) network is shown to be related to the Learning Vector Quantization algorithm. Based on this similarity, a hybrid training scheme for the RBF network is proposed. The resulting Rapid Kernel Classifier is evaluated using a 6-class radar data set. Considerable speedup in training is obtained with this new scheme. Also, for the one-dimensional case, we prove that the distribution of centroids of the RBF network ap proaches node density of the Self-Organizing Feature Map as a limit. This result suggests a deeper connection between the fundamental learning paradigms, namely supervised and unsupervised learning.
Get full access to this article
View all access options for this article.
References
1.
Beck, S. and J. Ghosh.1992. "Noise Sensitivity of Static Neural Classifiers" , m SPIE Conf. on Science of Artificial Neural Networks SPIE Proc. Vol. 1709 , Orlando, FL, April 1992.
2.
Broomhead, D.S. and D. Lowe.1988. "Multivariable Functional Interpolation and Adaptive Networks" , Complex Systems, 2:321-355.
3.
Chakravarthy, S., J. Ghosh, L. Deuser and S. Beck.1991. "Efficient Training Procedures for Adaptive Kernel Classifiers" , in Neural Networks for Signal Processing , pp. 21-29, IEEE Press.
4.
Geman, S., E. Bienenstock and R. Doursat.1992. "Neural Networks and the Bias/Vanance Dilemma" , Neural Computation, 4(1):1-58.
5.
Hartman, E. and J.D. Keeler.1991. "Predicting the Future: Advantages of Semi-Local Units" , Neural Computation, 3(4):566-578.
6.
LeCun, Y., J.S. Denker and S.A. Solla.1990. "Optimal Brain Damage", in Advances in Neural Information Processing Systems II, Denver, pp. 598-605.
7.
Huang, W.Y. and R.P. Lippmann.1987. "Neural Network and Traditional Classifiers", Neural Information Processing Systems, pp. 387-396.
Kowalski, J. , E. Hartman and J. Keeler.1990. "Layered Neural Networks with Gaussian Hidden Units as Umversal Approximators'; Neural Computation , 2:210-215.
10.
Kohonen, T.1989. Self- Organization and Associative MemoryBerlin: Springer-Verlag, 3rd Ed.
11.
Linde, Y., A. Buzo and R.M. Gray.1980. "An Algorithm for Vector Quantizer Design", IEEE Transactions on Communications, 28(1):84-95.
12.
Lee, S. and R.M. Kil.1988. "Multilayer Feedforward Potential Function Network" , in Proceedings of the Second International Conference on Neural Networks, pp. 161-171.
13.
Moody, J. and C.J. Darken.1989. "Fast Learning in Networks of Locally-Tuned Processing Units", Neural Computation, 1(2):281-294.
14.
Martinez, T., H.J. Ritter and K.J. Schulten.1990. "Three-Dimensional Neural Net for Learning Visuomotor Coordination of a Robot Arm", IEEE Trans. Neural Networks , 1:131-136.
15.
Poggio, T. and F. Girosi.1990. "Networks for Approximation and Learning", Proc. IEEE, 78(9):1481-1497.
16.
Park, J. and I.W. Sandberg.1991. "Universal Approximation Using Radial Basis Function Networks", Neural Computation, 3(2):246-257 (summer).
17.
Richard, M.D. and R.P. Lippmann.1991. "Neural Network Classifiers Estimate Bayesian a postenori Probabilities", Neural Computation , 3(4).461-483.
18.
Ritter, H.J., T. Martinez and K.J. Schulten.1992Neural Computation and Self-Organizing MapsAddison-Wesley Publishing Company.
19.
Ritter, H. and K. Schulten.1986. "On the Stationary State of Kohonen's Self-Organizing Sensory Mapping", Biological Cybernetics , 5499-106.
20.
Shoemaker, P.A., M.J. Carlin, R.L. Shimabukuro and C.E. Priebe.1991"Least Squares Learning and Approximation of Posterior Probabilities on Classification Problems by Neural Network Models", in Proc. 2nd Workshop on Neural Networks, WNN-AIND91, Auburn, pp. 187-196, February.
21.
Widrow, B. and M.E. Hoff.1960. "Adaptive Switching Circuits", 1960 IRE WESCON Convention Record, New York.IRE, pp. 96-104.
