K fold validation for feedforward net

Greetings

I have created a forwardnet and becuase I only have a handful of cases i need to use K fold validation. However, I am a bit confused on how to do that. I have seen crossval and cvpartition functions but i still do not know how to use them. can you help me train the network using kfold validation?

kind regards

Answer 1

[Gaurav Garg]

Hi Abdulaziz,

cvparition partitions data for cross-validation. It defines a random partition on data set and uses it to define training and test sets for validating a statistical model.

E.g. :

load ionosphere
tbl = array2table(X);
tbl.Y = Y;
rng('default') % For reproducibility
n = length(tbl.Y);
hpartition = cvpartition(n,'Holdout',0.3); % Nonstratified partition

hpartition holds the total number of observations, total number of test sets, the size of training set and the size of test size.

To display the indeces which are being used for training, you can use -

training(hpartition)

To display the indeces which are being used for testing, you can use -

test(hpartition)

While, using crossval, you can estimate the loss returned by 10-fold cross-validation error estimate.

err = crossval(criterion,X,y,'Predfun',predfun)
% returns a 10-fold cross-validation error estimate for the 
% function predfun based on the specified criterion

To know more about them, you can use the links provided - 1 2

[Abdulaziz Almershed]

but is there a buit in function to train a network using k-fold validation as validating method?