How to regenerate the regression plots from a stored network?

Hello all,

I have stored my trained network as .mat file. When i load the network, i can find all the information about the net. Can someone please tell me how can i obtain the plot performance and the regression plots that was generated first time after training.

Answer 1

[Shivam Singh]

It depends on how you have saved your network and what information it contains. If you have stored the network information in .mat file, then you get the information of training accuracy, training loss, validation accuracy etc., and then you can plot then to see the performance.

Example:

[net,info] = trainNetwork(imdsTrain,layers,options);
save('Output.mat','net','info');

In the above example, you have saved the “info” structure variable, which contains various training information.

But if you want to get the exact figure which pops when trainNetworkfunction is used, then you can try below commands when saving your network.

[net,info] = trainNetwork(___)
currentfig = findall(groot,'Tag','NNET_CNN_TRAININGPLOT_UIFIGURE');
savefig(currentfig,'test.fig');

The required figure will be saved as “test.fig”.

For more information regarding the function used above, you can refer below links:

trainNetwork, save, findall, savefig