which is the best way to select the input parameters for neural network
1 Ansicht (letzte 30 Tage)
Greg Heath am 11 Mai 2015
It depends on how much time and how many variables you have.
1. Standardize all variables. Then removing a variable is equivalent to replacing all of the values by it's mean value ... zero!
2. Minimize the number of hidden nodes to reduce complexity
3. The quickest way is to use MATLAB's backward stepwise selection routine for a linear model.
4. A more to the point method is to use backward selection with the minimum H net. Then you have several other options to consider
a. Retrain the net after the removal of each variable
b. Reconsider choosing previously removed variables
Hope this helps.
Thank you for formally accepting my answer