Hi Kenneth,
You can make use of Autoencoder object containing an autoencoder network, which consists of an encoder and a decoder.
- To create an Autoencoder, you may use trainAutoEncoder(X), which returns an autoencoder trained using the training data X.
- To specify the parameters that trainAutoEncoder uses for training the autoencoder, you may use TrainingParameters property.
- For encoding, you may use encode(autoenc,Xnew) which returns the encoded data for the input data Xnew, using the autoencoder, autoenc.
- For decoding, you may use decode(autoenc,Z) which returns the decoded data using the autoencoder, autoenc.
You can define custom architecture of autoencoder using deep learning layers. You can refer to this documentation for the list of deep learning layers supported in MATLAB. For example, a convolutional layer can be created as
layer = convolution2dLayer(11,96,'Stride',4)
% Includes a convolutional layer in a layers array
layers = [
imageInputLayer([28 28 1])
convolution2dLayer(5,20)
reluLayer
maxPooling2dLayer(2,'Stride',2)
fullyConnectedLayer(10)
softmaxLayer
classificationLayer];
You can use 2D / 3D conv layer/ any other layer as per your architecture. After defining the network, you can train the model. You can refer to this documentation .
Hope it helps!
