[Master Index] [Index for export/lyngby]

lyngby_nn_pruneobd

(export/lyngby/lyngby_nn_pruneobd.m)


Function Synopsis

[VMask,WMask] = lyngby_nn_pruneobd(V, W, ddCV, ddCW, arg1, ...

Help text

 lyngby_nn_pruneobd   - Pruning by Optimal Brain Damage

	function [VMask,WMask] = lyngby_nn_pruneobd(V, W, ddCV,ddCW,
           'PropertyName', 'PropertyValue')

	Input:	V      Input weights
               W      Output weights
               ddCV   Second order derivative of the costfunction with
                      respect to the input weights, matrix
                      conataining the diagonal of the hessian
               ddCW   Second order derivative of the costfunction
                      with respect to the output weights, matrix
                      containing the diagonal of the hessian
               Property:
                  'DeleteFraction'   {0.05} Fraction to delete 
                                     of remaining weights

	Output:	VMask   Mask for input weights. Deleted weight is
                       zero, else one
               WMask   Mask for output weights.

       Optimal Brain Damage pruning for a neural network.

       See also: lyngby_nn_cpruneobd, lyngby_nn_eprunobd,
                 lyngby_nn_qpruneobd

Cross-Reference Information

This function is called by

Produced by mat2html on Wed Jul 29 15:43:40 2009
Cross-Directory links are: OFF