Online selection of functional links for nonlinear system identification

This paper introduces a new method for improving nonlinear modeling performance in online learning by using functional link-based models. The proposed algorithm is capable of selecting the useful nonlinear elements resulting from the functional expansion, while setting to zero the ones that does not bring any improvement of the modeling performance. This allows to reduce any gradient noise due to a possible overestimate of the solution, thus preventing any overfitting phenomena. The proposed model is assessed in several nonlinear identification problems, including different levels of nonlinearity, showing significant improvements.