CMAC with General Basis Functions

Abstract The cerebellar model articulation controller (CMAC) is often used in learning control. It can be viewed as a basis function network (BFN). The conventional CMAC uses local constant basis functions. A disadvantage is that its output is constant within each quantized state and the derivative information is not preserved. If the constant basis functions are replaced by non-constant differentiable basis functions, the derivative information will be able to be stored into the structure as well. In this paper, the generalized scheme that uses general basis functions is investigated. The conventional CMAC is a special case of the generalized technique. The mathematical foundation for the modified scheme is derived and the convergence of learning is proved. Simulations for the CMAC with Gaussian basis functions (GBFs) are performed to demonstrate the improvement of accuracy in modeling, and the capability in providing derivative information. Copyright © 1996 Elsevier Science Ltd

[1]  Chun-Shin Lin,et al.  Learning convergence of CMAC technique , 1997, IEEE Trans. Neural Networks.

[2]  A. Sideris,et al.  Learning convergence in the cerebellar model articulation controller , 1992, IEEE Trans. Neural Networks.

[3]  D.A. Handelman,et al.  Theory and development of higher-order CMAC neural networks , 1992, IEEE Control Systems.

[4]  P. C. Parks.  Convergence Properties of Associative Memory Storage for Learning Control Systems , 1992 .

[5]  Qinghua Zhang,et al.  Wavelet networks , 1992, IEEE Trans. Neural Networks.

[6]  D. Broomhead,et al.  Radial Basis Functions, Multi-Variable Functional Interpolation and Adaptive Networks , 1988 .

[7]  W. T. Miller,et al.  CMAC: an associative neural network alternative to backpropagation , 1990, Proc. IEEE.

[8]  W. T. Miller,et al.  An overview of the CMAC neural network , 1991, [1991 Proceedings] IEEE Conference on Neural Networks for Ocean Engineering.

[9]  James S. Albus,et al.  New Approach to Manipulator Control: The Cerebellar Model Articulation Controller (CMAC)1 , 1975 .

[10]  P. C. Parks,et al.  Convergence Properties of Associative Memory Storage for Learning Control Systems , 1990 .

[11]  James S. Albus,et al.  Data Storage in the Cerebellar Model Articulation Controller (CMAC) , 1975 .

[12]  Sukhan Lee,et al.  Control of nonlinear system with input saturation based on Gaussian potential function network , 1991, [Proceedings] 1991 IEEE International Joint Conference on Neural Networks.

[13]  Sukhan Lee,et al.  A Gaussian potential function network with hierarchically self-organizing learning , 1991, Neural Networks.