from models import gaussian_mnist, fit_vae
from data.my_mnist import x_train, x_test
from visualize import render_grid
# Create and train the VAE model
vae = gaussian_mnist('vae', latent_dim=2, pixel_std=.05, k=8)
fit_vae(vae, x_train, x_test, epochs=100, weights_file='vae-weights.h5')
# Visualize results
render_grid(vae.latent.flat_samples, vae.reconstruction)
# Create and train the IWAE model
vae = gaussian_mnist('iwae', latent_dim=2, pixel_std=.05, k=8)
fit_vae(vae, x_train, x_test, epochs=100, weights_file='iwae-weights.h5')
# Visualize results
render_grid(vae.latent.flat_samples, vae.reconstruction)
from models import gaussian_mnist, fit_vae from data.my_mnist import x_train, x_test from visualize import render_grid # Create and train the model vae = gaussian_mnist(latent_dim=2, pixel_std=.05) fit_vae(vae, x_train, x_test, epochs=100, weights_file='weights.h5') # Visualize results render_grid(vae.latent.sample, vae.reconstruction)
mnist_surface = pygame.Surface((28, 28))
np_mnist_surface_view = np.frombuffer(mnist_surface.get_buffer())
mnist_argb = np.zeros((28, 28, 4), dtype=np.uint8)
ent_surface = pygame.Surface((100, 100))
np_ent_surface_view = np.frombuffer(ent_surface.get_buffer())
class_argb = np.zeros((100, 100, 4), dtype=np.uint8)
game_exit = False
clock = pygame.time.Clock()
mouse_position = (0, 0)
model_type = argv[1]
latent_dim = 2
pixel_std = .05
vae = gaussian_mnist(model_type,
latent_dim=latent_dim,
pixel_std=pixel_std,
k=1)
weights_file = os.path.join("models", "mnist_%s.h5" % model_type)
vae.model.load_weights(weights_file)
#####################
# GET ENTROPY IMAGE #
#####################
# Get color for each numeric class
class_colors = get_class_colors(10)
# Get keras function of recognition model.
q = K.function([vae.inpt], [vae.latents[0].mean, vae.latents[0].log_var])
# Apply recognition model to test set.