A SLIDING MODE STRATEGY FOR ADAPTIVE LEARNING IN ADALINES

A dynamical sliding mode control approach is proposed for robust adapt ive learning in analog Adaptive Linear Elements (Adalines), constituti ng basic building blocks for perceptron-based feedforward neural netwo rks, The zero level set of the learning error variable is regarded as a sliding surface in the space of learning parameters, A sliding mode trajectory can then be induced, in finite time, on such a desired slid ing manifold, Neuron weights adaptation trajectories are shown to be o f continuous nature, thus avoiding bang-bang weight adaptation procedu res, Sliding mode invariance conditions determine a least squares char acterization of the adaptive weights average dynamics whose stability features may be studied using standard time-varying linear systems res ults, Robustness of the adaptative learning algorithm, with respect to bounded external perturbation signals, and measurement noises, is als o demonstrated, The article presents some simulation examples dealing with applications of the proposed algorithm to forward and inverse pla nt dynamics identification.