My new question is about activation function for the output layer in regression model in Neural Network SO: Which is better? Linear activation function or NO activation func