The CMAC (Cerebellar Model Articulation Controller) suffers from two important problems: the huge amount of memory needed for its implementation in many common situations, and the lack of a systematic way for selecting appropriate values for its parameters, particularly number of quantization intervals. This paper presents two proposals for addressing these difficulties: 1) a dynamic implementation that requires memory only for those weights needed to represent the training data set, and that performs linear interpolation when a query using other weights is requested; and 2) consists of the definition of an index of correlation from which the optimal number of quantization intervals that should be assigned to each dimension of the input space that can be found. Experiments are performed for two synthetic cases and for one set of real data. These are used to model the dynamic behaviour of a real sensor-based car. Figures are given to show the memory savings and mean squared error obtained.
[1]
Antonio Artés-Rodríguez,et al.
Fourier analysis of the generalized CMAC neural network
,
1998,
Neural Networks.
[2]
James S. Albus,et al.
New Approach to Manipulator Control: The Cerebellar Model Articulation Controller (CMAC)1
,
1975
.
[3]
Frank L. Lewis,et al.
CMAC neural networks for control of nonlinear dynamical systems: Structure, stability and passivity
,
1997,
Autom..
[4]
Chun-Shin Lin,et al.
CMAC with General Basis Functions
,
1996,
Neural Networks.
[5]
Sabri Cetinkunt,et al.
CMAC Neural Network Control for High Precision Motion Control in the Presence of Large Friction
,
1995
.