MATLAB Answers

Can we only train the classification layer when do transfer learning of a pre-trained network?

9 views (last 30 days)
The question is, can we only train the classification layer when do transfer learning of a pre-trained network? I want to speed up my training by keeping the feature extraction layers (base model) as they are and only replace and retrain the classification layers.
The equivalent way in Keras (Python) is by: base_model.trainable = False
If possible in Matlab, please let me know how. Your help is appreciated.


Greg Heath
Greg Heath on 17 Jul 2020
See if it is now possible to assign different learning rates to the different layers. I wasn't able to some time ago.
Sud Sudirman
Sud Sudirman on 17 Jul 2020
What we can do it to set the InitialLearnRate to a very small number e.g., 1e-4 and set the WeightLearnRateFactor and BiasLearnRateFactor of the FCNN before the last Classification Layer to a large number (e.g., 20) as was demonstrated here . This approach seems to approximate the approach you suggested, Greg.
However, I was thinking about something more in the line of what described in In this approach, I can record the activations of the last layer (bottleneck features) before the FCNN and use them to train a classifier. But for some reason, I cannot use these features to train an network containing {SequenceInputLayer+FCNN+ClassificationLayer}. Odd.

Sign in to comment.

Accepted Answer

Srivardhan Gadila
Srivardhan Gadila on 17 Jul 2020
In order to freeze the weights of a particular layer of your network set the properties WeightLearnRateFactor & BiasLearnRateFactor to zero. Refer to fullyconnectedLayer - Learn Rate and Regularization, convolution2dLayer - Learn Rate and Regularization & lstmLayer - Learn Rate and Regularization.
layer.WeightLearnRateFactor = 0;

More Answers (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!