def generative_adversarial_network_example():
ed.set_seed(42)
data_dir = '/tmp/data'
out_dir = '/tmp/out'
if not os.path.exists(out_dir):
os.makedirs(out_dir)
M = 128 # Batch size during training.
d = 100 # Latent dimension.
(x_train, _), (x_test, _) = mnist(data_dir)
x_train_generator = generator(x_train, M)
x_ph = tf.placeholder(tf.float32, [M, 784])
#--------------------
# GANs posit generative models using an implicit mechanism.
# Given some random noise, the data is assumed to be generated by a deterministic function of that noise.
with tf.variable_scope('Gen'):
eps = Uniform(tf.zeros([M, d]) - 1.0, tf.ones([M, d]))
x = generative_network(eps)
#--------------------
# In Edward, the GAN algorithm (GANInference) simply takes the implicit density model on x as input, binded to its realizations x_ph.
# In addition, a parameterized function discriminator is provided to distinguish their samples.
inference = ed.GANInference(data={x: x_ph}, discriminator=discriminative_network)
# We'll use ADAM as optimizers for both the generator and discriminator.
# We'll run the algorithm for 15,000 iterations and print progress every 1,000 iterations.
optimizer = tf.train.AdamOptimizer()
optimizer_d = tf.train.AdamOptimizer()
inference = ed.GANInference(data={x: x_ph}, discriminator=discriminative_network)
inference.initialize(optimizer=optimizer, optimizer_d=optimizer_d, n_iter=15000, n_print=1000)
# We now form the main loop which trains the GAN.
# At each iteration, it takes a minibatch and updates the parameters according to the algorithm.
sess = ed.get_session()
tf.global_variables_initializer().run()
idx = np.random.randint(M, size=16)
i = 0
for t in range(inference.n_iter):
if t % inference.n_print == 0:
samples = sess.run(x)
samples = samples[idx, ]
fig = plot(samples)
plt.savefig(os.path.join(out_dir, '{}.png').format(str(i).zfill(3)), bbox_inches='tight')
plt.close(fig)
i += 1
x_batch = next(x_train_generator)
info_dict = inference.update(feed_dict={x_ph: x_batch})
inference.print_progress(info_dict)
def test_gan_inference(self):
N, D, W_1, W_2, W_3, b_1, b_2, x_ph, y, X_train, y_train = self._test()
with tf.variable_scope("Gen"):
theta = tf.get_variable("theta", [1])
y = tf.cast(y, tf.float32) * theta
def discriminator(x):
w = tf.get_variable("w", [1])
return w * tf.cast(x, tf.float32)
inference = ed.GANInference(data={
y: tf.cast(y_train, tf.float32),
x_ph: X_train
},
discriminator=discriminator)
inference.run(n_iter=1)
def main(_):
sns.set(color_codes=True)
ed.set_seed(42)
# DATA. We use a placeholder to represent a minibatch. During
# inference, we generate data on the fly and feed `x_ph`.
x_ph = tf.placeholder(tf.float32, [FLAGS.M, 1])
# MODEL
with tf.variable_scope("Gen"):
eps = tf.linspace(-8.0, 8.0,
FLAGS.M) + 0.01 * tf.random_normal([FLAGS.M])
eps = tf.reshape(eps, [-1, 1])
x = generative_network(eps)
# INFERENCE
optimizer = tf.train.GradientDescentOptimizer(0.03)
optimizer_d = tf.train.GradientDescentOptimizer(0.03)
inference = ed.GANInference(data={x: x_ph},
discriminator=discriminative_network)
inference.initialize(optimizer=optimizer, optimizer_d=optimizer_d)
tf.global_variables_initializer().run()
for _ in range(inference.n_iter):
x_data = next_batch(FLAGS.M).reshape([FLAGS.M, 1])
info_dict = inference.update(feed_dict={x_ph: x_data})
inference.print_progress(info_dict)
# CRITICISM
db, pd, pg = get_samples(x_ph)
db_x = np.linspace(-8, 8, len(db))
p_x = np.linspace(-8, 8, len(pd))
f, ax = plt.subplots(1)
ax.plot(db_x, db, label="Decision boundary")
ax.set_ylim(0, 1)
plt.plot(p_x, pd, label="Real data")
plt.plot(p_x, pg, label="Generated data")
plt.title("1D Generative Adversarial Network")
plt.xlabel("Data values")
plt.ylabel("Probability density")
plt.legend()
plt.show()
def test_normal(self):
with self.test_session() as sess:
# DATA
M = 12 # batch size during training
x_ph = tf.placeholder(tf.float32, [M, 1])
# MODEL
with tf.variable_scope("Gen"):
theta = tf.Variable(0.0)
x = Normal(theta, 0.1, sample_shape=[M, 1])
# INFERENCE
inference = ed.GANInference(
data={x: x_ph}, discriminator=discriminative_network)
inference.initialize(n_iter=1000)
tf.global_variables_initializer().run()
for _ in range(inference.n_iter):
x_data = next_batch(M).reshape([M, 1])
inference.update(feed_dict={x_ph: x_data})
# CRITICISM
self.assertAllClose(theta.eval(), 4.0, rtol=1.0, atol=1.0)
os.makedirs(IMG_DIR)
# DATA. MNIST batches are fed at training time.
mnist = input_data.read_data_sets(DATA_DIR, one_hot=True)
x_ph = tf.placeholder(tf.float32, [M, 784])
# MODEL
with tf.variable_scope("Gen"):
eps = Uniform(a=tf.zeros([M, d]) - 1.0, b=tf.ones([M, d]))
x = generative_network(eps)
# INFERENCE
optimizer = tf.train.AdamOptimizer()
optimizer_d = tf.train.AdamOptimizer()
inference = ed.GANInference(data={x: x_ph},
discriminator=discriminative_network)
inference.initialize(optimizer=optimizer,
optimizer_d=optimizer_d,
n_iter=15000,
n_print=1000)
sess = ed.get_session()
tf.global_variables_initializer().run()
idx = np.random.randint(M, size=16)
i = 0
for t in range(inference.n_iter):
if t % inference.n_print == 0:
samples = sess.run(x)
samples = samples[idx, ]