How to interpret the performance graph in NN Toolbox ?
5 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Ashik Joshy
am 22 Jun. 2015
Beantwortet: Greg Heath
am 10 Jul. 2015
I have created a neural network with 53 inputs and 7 outputs with 10 hidden layers. I need the performance graph with MSE vs Epochs for training only. What is the meaning of all three graphs showing there. ? I did not understand the red graph and the green one. How to plot MSE vs Epochs for training ? the blue one alone. Somebody please help....
0 Kommentare
Akzeptierte Antwort
Greg Heath
am 10 Jul. 2015
The data division indices are stored in the structure tr ...
[ net tr y e ] = train(net, x, t );
Data is automatically divided into
training data used to calculate the weights
nontraining validation data used to help make sure that the performance trend of nontraining data is improvement. If not for 6 straight epochs, training will be stopped. (green curve)
nontraining test data used to obtain an unbiased estimate of performance on unseen data. (red curve)
To plot only the training data
plot( y(tr.trainInd) );
However, the training data is highly biased, especiall if the number of training equations is not much larger than the number of unknown weights.
That is why supervisors and clients only care about performance on test data.
Hope this helps.
Greg
0 Kommentare
Weitere Antworten (0)
Siehe auch
Kategorien
Mehr zu Sequence and Numeric Feature Data Workflows finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!