Neural Networks for Regression Problems

The hidden layer squash function, ϕ h , that is used by JMP is the hyperbolic tangent function and I believe nnet in R uses the logistic activation function for the hidden layers.For regression problems, it is common to include a skip-layer to the neural network. Also for regression problems it is important that the final outputs be linear as we don’t want to constrain the predictions to be ... ................
................