Training of supervised neural networks via a nonlinear primal-dual interior-point method

Nicolas Couellan (1, 2) , T.B. Trafalis (3) , S.C. Bertrand (3)

1 ENAC - Ecole Nationale de l'Aviation Civile
2 IMT - Institut de Mathématiques de Toulouse UMR5219
3 OU - University of Oklahoma

We propose a new training algorithm for feedforward supervised neural networks based on a primal-dual interior-point method for nonlinear programming. Specifically, we consider a one-hidden layer network architecture where the error function is defined by the L/sub 2/ norm and the activation function of the hidden and output neurons is nonlinear. Computational results are given for odd parity problems with 2, 3, and 5 inputs respectively. Approximation of a nonlinear dynamical system is also discussed.

Domaines: Optimisation et contrôle [math.OC]; Apprentissage [cs.LG]

Nicolas Couellan, T.B. Trafalis, S.C. Bertrand. Training of supervised neural networks via a nonlinear primal-dual interior-point method. International Conference on Neural Networks (ICNN'97), Jun 1997, Houston, United States. ⟨10.1109/ICNN.1997.614210⟩. ⟨hal-01922685⟩