I used Matlab to train the network with 3-18-8. I used Matlab default and I am not sure that it is right to use the Levenberg-Marquardt back-propagation as learning method for MLP training.
I am appreciated if some one could help me to understand.
Levenberg-Marquardt (trianlm) is an algorithm that uses backpropogation. There is other algorithms that uses backpropogation such as Bayesian Regularization (trainbr) and Scaled Conjugate Gradient (trainscg).
Point 9 in link below briefly describe the difference: