Improved generalization learning with sliding mode control and the Levenberg-Marquadt algorithm

A variation of the well known Levenberg-Marquardt for training neural networks is presented in this work. The algorithm presented restricts the norm of the weights vector to a preestablished norm value and finds the minimum error solution for that norm value. A range of different norm solutions is generated and the best generalization solution is selected. The results show the efficiency of the algorithm in terms of convergence speed and generalization performance.