Shape Control Learning Algorithm for Neural Networks

Abstract:

A new kind of shape control learning algorithm (SCLA) for training neural networks is proposed. We use the rational cubic spline (with quadratic denominator) to implement a new neural system for shape control, and construct a new kind of artificial neural networks based on given patterns. The shape can be controlled by some shape parameters, which is much different from the known algorithms for training neural networks. The numerical experiments indicate that the new method proposed in this paper demonstrates good results.

• Periodical: Applied Mechanics and Materials (Volumes 347-350)
• Pages: 2270-2274
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.347-350.2270
• Online since: August 2013
• Authors: Dai Yuan Zhang
• Keywords: Neural Network (NN), Rational Spline, Shape Control, Weight Function

References

[1] D. Y. Zhang, New Theories and Methods on Neural Networks, Beijing: Tsinghua University Press, 2006 (in Chinese).

[2] D. Y. Zhang, New algorithm for training feedforward neural networks with cubic spline weight functions, Systems Engineering and Electronics, Vol. 28, Sep. 2006, pp.1434-1437 (in Chinese).

[3] D. Y. Zhang, New algorithm for training neural networks based on generalized Чебышев polynomials, Systems Engineering and Electronics, Vol. 30, Nov. 2008, pp.2274-2279 (in Chinese).

[4] D. Y. Zhang, A new algorithm of neural networks with B-Spline weight functions, Proc. 2010 International Conference on Artifical Intelligence and Education. Hangzhou: IEEE Press, 2010, pp.782-785.

[5] R. J. Ning, G. Q. Zhu, The construction of a kind of rational spline, Numerical Mathematics: A Journal of Chinese Universities, Vol. 2, 1996, pp.134-138.

[6] F. N. Fritsch, R. E. Carlson. Monotone piecewise cubic interpolation, SIAM Journal on Numerical Analysis, Vol. 17, Feb. 1980, pp.238-246.

DOI: 10.1137/0717021