How to extract feature from autoencoder/stackautoencoder ?
6 views (last 30 days)
Suppose, 2 autoencoder is stacked together and a softmax layer is used for classification.
At the time of fine tuning %deepnet = stack(autoenc1,autoenc2,softnet); [deepnet,tr]=train(deepnet,X,T);% are used.
After finetuning, i need feature from 'autoenc1'' or 'autoenc2'.
How can i extract the feature?