Neuronal Homotopy Regressors

Abstract

In this work, a neural model is proposed, which solves two limitations of artificial neural networks. The first refers to the ability to extrapolate outside the domain of the training data. The second arises when only a small sample is available for training. On the other hand, there is a need to characterize a complex system, and the dynamics of its components is partially known. The proposed model is based on the construction of a regressor from a feasible space, using Homotopy Analysis. In this way, a functional neural network is obtained.