Evaluation network performance using cross-validation

by Björn Lindqvist   Last Updated March 20, 2019 17:19 PM

Suppose I have a data set on which I'm training a neural network. I'm using four-fold validation, meaning that I train four models, one for each fold. Two of the folds are used for training, one for testing and one for validation so that no model is trained on its testing data.

The result of this process is four trained models (obviously). My question is what is the correct way of reporting the performance for my network? Should I report the arithmetic mean of these four models on their respective folds or something else?

Related Questions

Do I include the validation set in final training?

Updated January 22, 2019 22:19 PM

Designing structure of CNN

Updated April 15, 2017 18:19 PM

Validate classification models

Updated January 25, 2019 07:19 AM