training progress graph of 'trainNetwork'

Hi,

I am using trainNetwork function in deep neural network toolbox.

A graph appears automatically appears showing training progress. I like to know how can the plot styles (say, color, label size) be modified by user?

Also it is of interest to save the plot for future reference. But unlike matlab figure window, it doesnot allow that also.

Pls inform me on how to access and modify settings in training progress graph.

Thanks,

Deepak

Answer 1

[Areej Varamban Kallan]

Hi Deepak,

Please try the following:

1. After the training is complete, close all the figures except the training plot.
2. Execute the following commands on MATLAB command line:

    h= findall(groot,'Type','Figure');
    h.MenuBar = 'figure';

The training plot will now have a menu bar.

3. Click Edit-> Figure Properties. This will open the Poperty Editor window.

4. Use the options in the Property Editor window to change plot styles.

5. Similary, you can save the figure from 'File->Save'

To hide the menu bar, execute the following command on MATLAB command line:

    h.MenuBar = 'none';

I hope this helps.

[Deepak Sivadas]

Please address the issues below,

h= findall(groot,'Type','Figure ');

creates --- 0x0 graphicsPlaceholder and

h.MenuBar = 'figure ';

returns error message : Property assignment is not allowed when the object is empty. Use subscripted assignment to create an array element.

[Areej Varamban Kallan]

Hi Deepak,

Please make sure that the figure is open. Which version of MATLAB are you using?

[Deepak Sivadas]

Hi Areej,

how to save it as fig window? Your earlier suggestions did not work.

[Matlab 2018a]

[z xu]

Great! It's helpful. Thanks buddy !👍

[Asif Mehmood]

@Areej Varamban Kallan, Thanks!

Your method worked for me, and I am using MATLAB 2020a.

[nedim moglu]

Hi,@Areej Varamban Kallan,

h= findall(groot,'Type','Figure');
 h.MenuBar = 'figure';

while only Deep Learning Training Progress figure is open I used advised codes.after

>>

h.MenuBar = 'figure';

I get the error like this:

Functionality not supported with figures created with the uifigure function. For more

information, see Graphics Support in App Designer.

Do you have any idea about this error.

I have Matlab 2020b

[Fatma Diab]

Hi, @nedim moglu

I have the same problem. Can you help me to solve this error please?

Functionality not supported with figures created with the uifigure function.

[Towfeeq Fairooz]

It would have been easier if that worked as per the help provided by some user. Check the video (MATLAB 2020a)

https://www.youtube.com/watch?v=U183m4Brtyg

But looks like it does not work the same way in MATLAB 2020b and it gives error : 'Functionality not supported with figures created with the uifigure function'.

uifigure MenuBar is not yet enabled in this version . @MathWorks Support Team can you please fix that. It would be easier to export the figures and can make life easier. Thank you.

in my application, I used exportapp function :

[net,info]=trainNetwork(...);
currentfig = findall(groot, 'Tag', 'my_CNN_trainingPlot_uifigure') ; 
%savefig(currentfig,'CNN_trainingplot1.fig'); 
exportapp(currentfig,'CNN_tainingplot2.pdf');

it worked for me. Hope it helps

Can export as '.png' file too.

[kh rezaee]

I appreciate your response, Areej Varamban Kallan. In the 2020 version, I was able to edit the output after I used your command for training progress figure. However, I am unable to edit the figure in version 2022, and I receive the following error message. Is there a way to make the menu appear?

Functionality not supported with figures created with the uifigure function.

[Qiaohui He]

Maybe u can try

[net tr] = trainNetwork(XTrain,YTrain,layers,options);

save(tr)

and the 'tr'will store the data of the training process. Then the data in workspace of tr.mat could be applied to use plot function.

Answer 2

[avi.loey]

Thankyou. This was exactly what I needed. Works perfectely !!!!

Answer 3

[Amit DOegar]

h= findall(groot,'Type','Figure');

h.MenuBar = 'figure';

How to do this for multiple open figures as in k-fold we need to save all plots

or how to get handle of all open figures in matlab

[krishna Chauhan]

Hey Amit

How are you able to see what cross validation its training?

I mean where to assign cross k validation in trainNetwork?

Answer 4

[Xiao Liu]

you can use the data from working space and plot one more picture for yourself.