Training of supervised neural networks via a nonlinear primal-dual interior-point method
Couellan, Nicolas ; Trafalis, T.B. ; Bertrand, S.C.
HAL, hal-01922685 / Harvested from HAL
Text on HAL
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.
Publié le : 1997-06-04
Classification:  [MATH.MATH-OC]Mathematics [math]/Optimization and Control [math.OC],  [INFO.INFO-LG]Computer Science [cs]/Machine Learning [cs.LG] 
@article{hal-01922685,
     author = {Couellan, Nicolas and Trafalis, T.B. and Bertrand, S.C.},
     title = {Training of supervised neural networks via a nonlinear primal-dual interior-point method},
     journal = {HAL},
     volume = {1997},
     number = {0},
     year = {1997},
     language = {en},
     url = {http://dml.mathdoc.fr/item/hal-01922685}
}

Couellan, Nicolas; Trafalis, T.B.; Bertrand, S.C. Training of supervised neural networks via a nonlinear primal-dual interior-point method. HAL, Tome 1997 (1997) no. 0, . http://gdmltest.u-ga.fr/item/hal-01922685/