Hands-On Deep Learning Algorithms with Python

Defining the sampling operation

Define the sampling operation with a reparameterization trick that samples the latent vector from the encoder's distribution:

def sampling(args):    z_mean, z_log_var = args    epsilon = K.random_normal(shape=(K.shape(z_mean)[0], latent_dim), mean=0., stddev=epsilon_std)    return z_mean + K.exp(z_log_var / 2) * epsilon

Sample the latent vector z from the mean and variance:

z = Lambda(sampling, output_shape=(latent_dim,))([z_mean, z_log_var])