On the selection of nodes in linear-in-the-weight neural networks

In this paper, we propose algorithms for selecting the regressor terms in linear-in-the-weight neural networks. These algorithms are accompanied by appropriate learning algorithms for adjusting the weights of the neural network. By analyzing an appropriate error functional, we investigate the convergence properties of the proposed algorithms; moreover, we investigate the optimality of these algorithms and we construct conditions-regarding the nature of the regressor terms-under which the proposed algorithms are optimal.