Neural networks organizations to learn complex robotic functions

This paper considers the general problem of function es- timation with a modular approach of neural computing. We propose to use functionally independent subnetworks to learn complex functions. Thus, function approximation is decomposed and amounts to estimate different elementary sub-functions rather than the whole function with a single network. This modular decomposition is a way to introduce some a priori knowledge in neural estimation. Functionally independent subnet- works are obtained with a bidirectional learning scheme. Implemented with self-organizing maps, the modular approach has been applied to a robot control problem, a robot positioning task.

[1]  Geoffrey E. Hinton,et al.  Adaptive Mixtures of Local Experts , 1991, Neural Computation.

[2]  Patrice Wira,et al.  Multiple self-organizing maps to facilitate the learning of visuo-motor correlations , 1999, IEEE SMC'99 Conference Proceedings. 1999 IEEE International Conference on Systems, Man, and Cybernetics (Cat. No.99CH37028).

[3]  Mitsuo Kawato,et al.  MOSAIC Model for Sensorimotor Learning and Control , 2001, Neural Computation.

[4]  Ling Guan,et al.  Modularity in neural computing , 1999, Proc. IEEE.

[5]  Yasuharu Koike,et al.  PII: S0893-6080(96)00043-3 , 1997 .

[6]  Jean-Luc Buessler,et al.  Additive Composition of Supervised Self-Organizing Maps , 2002, Neural Processing Letters.