Stacked autoencoder in TensorFlow

The steps to build a stacked autoencoder model in TensorFlow are as follows:

1. First, define the hyper-parameters as follows:

learning_rate = 0.001
n_epochs = 20
batch_size = 100
n_batches = int(mnist.train.num_examples/batch_size)

2. Define the number of inputs (that is, features) and outputs (that is, targets). The number of outputs will be the same as the number of inputs:

# number of pixels in the MNIST image as number of inputs
n_inputs = 784
n_outputs = n_inputs

3. Define the placeholders for input and output images:

x = tf.placeholder(dtype=tf.float32, name="x", shape=[None, n_inputs])
y = tf.placeholder(dtype=tf.float32, name="y", shape=[None, n_outputs])

4. Add the number of neurons for encoder and decoder layers as [512,256,256,512]:

# number of hidden layers
n_layers = 2
# neurons in each hidden layer
n_neurons = [512,256]
...