I want to store the data of MSE vs epoch obtained during training of the neural network. By using "plotperform(TR)" we are able to get the MSE vs epoch graph but not the data itself. How to store this MSE vs epoch data in array variable?

how to obtain MSE vs Epoch data of training the neural network in MATLAB? %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

clc;clear all;close all;

t=0:0.01:10;
s=sin(2*pi*1*t);

in=t;
out=s;

net_size_array={[5] [10] [20] [5 5] [10 10] [20 20] [5 5 5] [10 10 10] [20 20 20]};
size_training=length(net_size_array);

for i=1:size_training
    net=fitnet(cell2mat(net_size_array(i)));
    [net, TR]=train(net,in,out);
    figure(1)
    plotperform_pk(TR);grid on;hold on;

end

%% I want to store and plot the MSE vs Epoch graph for various neural network size.

Answer 1

[Sai Praneet]

In your code the line:

[net, TR]=train(net,in,out);

stores all the essential information in the structure variable "TR". To get the Epoch vs. MSE data corresponding to the training, validation, and testing set all we have to do is store them in some variables for each network size such as

train_mse = TR.perf ;

val_mse = TR.vperf;

test_mse = TR.tperf;