defining divideblock function for feedforward net

i've used 'divideblock' function as follow:

net.divideFcn = 'divideblock'; net.divideParam.trainRatio = 0.6; net.divideParam.valRatio = 0.20; net.divideParam.testRatio = 0.20;

for feedforward neural network. when i run the program, the details of the network will be displayed. the problem is, it display 2 neural network details. the first set will show

net =

    Neural Network

. . . .

    functions:

          adaptFcn: 'adaptwb'
        adaptParam: (none)
          derivFcn: 'defaultderiv'
         divideFcn: 'dividerand'

and the other one shows

    functions:

          adaptFcn: 'adaptwb'
        adaptParam: (none)
          derivFcn: 'defaultderiv'
         divideFcn: 'divideblock'
       divideParam: .trainRatio, .valRatio, .testRatio

. . . .

is it suppose to be like this when we define 'divideblock' as a divide function? i expect it to display only 1 and not both, since i dont need my data to be randomized.

thank you. :)

[Greg Heath]

Post your code.

P.S. Use fitnet for regression and patternnet for classification. Both call feedforwardnet and provide better output info. There is no reason to use feedforwardnet.

Answer 1

[Greg Heath]

The default 'dividerand' exists at net creation: net = fitnet(H)

Specifying 'divideblock' then replaces it.