How can I get the 3 minimum values of this plot?

How can I get the number of neurons on which I get the smallest RMSE for Training, Validation and Test set.

for i = 1:60
    
    net = feedforwardnet(i);
    net.trainParam.epochs = 300;
    [net, tr] = train(net, input, output, 'useparallel', 'yes');
    
    pnet_train = net(input(:,tr.trainInd)); % prediction
    tnet_train = output(:,tr.trainInd);     % target
    rmse_train(i) = sqrt(mean((pnet_train - tnet_train).^2));
    
    
    pnet_vali = net(input(:,tr.valInd)); % prediction
    tnet_vali = output(:,tr.valInd);     % target
    rmse_vali(i) = sqrt(mean((pnet_vali - tnet_vali).^2));
    
    pnet_test = net(input(:,tr.testInd)); % prediction
    tnet_test = output(:,tr.testInd);     % target
    rmse_test(i) = sqrt(mean((pnet_test - tnet_test).^2));
    
end

figure;
plot(1:60,rmse_train,'LineWidth',2)
hold on;
plot(1:60,rmse_vali,'LineWidth',2)
hold on;
plot(1:60,rmse_test,'LineWidth',2)
title('Performance');
legend('Training','Validation','Test');
xlabel('Neurons');
ylabel('RMSE');
axis auto;
grid on;

[the cyclist]

What information do you have?

• Only the image?
• The figure file for the image?
• The data that was used to create the image?

[Bob]

Hi!

Thank you for responding.

I have added the code.

Answer 1

[KSSV]

REad about min. You have saved rmse of train, validatioon and testing. Use the function min and get the index. I would say check it for train and pick this index.

[val,idx] = min(rmse_train) ;
[idx rmse_train(idx) rmse_vali(idx) rmse_test(idx)]