How can I save and reload trained autoencoders?

I am trying to save a trained autoencoder so I can reload it at a another point in time and run it. When I try: save autoenc; It saves to a .mat file. When I try: load autoenc; It comes in as a mat object, not an autoencoder. What is the proper way to do this?

Answer 1

[Corey Silva]

I'm having trouble reproducing what you described. If I do the following, I see the autoencoder loaded just fine.

>> X = abalone_dataset;
>> autoenc = trainAutoencoder(X);
>> save autoenc
>> clear all
>> load autoenc

It's possible you're not using the load/save functions correctly in this context. It might be worth taking a look at the following pages: https://www.mathworks.com/help/matlab/ref/save.html https://www.mathworks.com/help/matlab/ref/load.html

[Paul Poloniewicz]

Thanks for your response! Here's what I get:

Trial>> autoenc autoenc = Autoencoder with properties:

                 HiddenSize: 25
    EncoderTransferFunction: 'logsig'
             EncoderWeights: [25×784 double]
              EncoderBiases: [25×1 double]
    DecoderTransferFunction: 'logsig'
             DecoderWeights: [784×25 double]
              DecoderBiases: [784×1 double]
         TrainingParameters: [1×1 struct]
                  ScaleData: 1
Trial>> TestAutoenc = autoenc
TestAutoenc = 
  Autoencoder with properties:

                 HiddenSize: 25
    EncoderTransferFunction: 'logsig'
             EncoderWeights: [25×784 double]
              EncoderBiases: [25×1 double]
    DecoderTransferFunction: 'logsig'
             DecoderWeights: [784×25 double]
              DecoderBiases: [784×1 double]
         TrainingParameters: [1×1 struct]
                  ScaleData: 1
Trial>> save 'SavedAutoenc' autoenc
Trial>> RestoredAutoenc = load('SavedAutoenc.mat')
RestoredAutoenc = 
  struct with fields:     ---(Not an AutoEncoder!)

    autoenc: [1×1 Autoencoder]
Trial>> view(RestoredAutoenc)
Error using view (line 73)   --- (no such method)
Invalid input arguments 
Trial>> view(TestAutoenc)    ---(this runs normally)
Trial>>