Consider the following code for a simple AutoEncoder, what is model_1 outputting ?

inputs = tf.keras.layers.Input(shape=(784,))

def simple_autoencoder():

encoder = tf.keras.layers.Dense(units=32, activation=’relu’)(inputs)

decoder = tf.keras.layers.Dense(units=784, activation=’sigmoid’)(encoder)

return encoder, decoder

output_1, output_2 = simple_autoencoder()

model_1 = tf.keras.Model(inputs=inputs, outputs=output_1)

model_2 = tf.keras.Model(inputs=inputs, outputs=output_2)

options:
Displaying the reconstruction of the original input which was fed to this architecture.

Displaying the classification layer of the model, mapping input to the output label.

Displaying the label value which the model is trying to reconstruct.

Displaying the internal representation of the input the model is learning to replicate.