ICP_Deep Learning 6 - PallaviArikatla/Python GitHub Wiki

INTRODUCTION: To analyze simple and stacked auto-encoders.

IMPLEMENTATION:

Question 1: Add one more hidden layer to autoencoder.

  • We will be creating a autoencoder model.
  • Add a hidden layer with "relu" activation and 512 neurons on the encoding side and apply this for the decoding side also.
  • Compile the X_test data and obtain the output.

Question 2: Do the prediction on the test data and then visualize one of the reconstructed version of that test data. Also, visualize the same test data before reconstruction using Matplotlib.

  • Perform prediction on the data after creating auto-encoder model.
  • Show/Plot the original and reconstructed data.

  • Plot for loss and accuracy.

Question 3: Repeat the question 2 on the denoisening auto-encoder.

  • Here we are introducing noise in the x_test data and fitting it to the auto-encoder and creating a history object with 10 epochs.

  • Input image.

  • Noisy input image.

  • Reconstructed image.

Question 4: Plot the loss and accuracy using history object.

  • Plots.

BONUS QUESTION: Visualize the compressed version of the input data in the middle layer.

  • Add the encode and decode layers.
  • Assign and compile the model.
  • Split the model to train and test.
  • Normalize the model.
  • Obtain predicted values and display them.