def test_mmd_name(self):
with self.test_session():
x = tf.random_uniform((2, 3), seed=1)
kernel = partial(utils.gaussian_kernel_matrix, sigmas=tf.constant([1.]))
loss = losses.maximum_mean_discrepancy(x, x, kernel)
self.assertEquals(loss.op.name, 'MaximumMeanDiscrepancy/value')
def test_mmd_is_zero_when_inputs_are_same(self):
with self.test_session():
x = tf.random_uniform((2, 3), seed=1)
kernel = partial(utils.gaussian_kernel_matrix,
sigmas=tf.constant([1.]))
self.assertEquals(
0,
losses.maximum_mean_discrepancy(x, x, kernel).eval())
def test_mmd_name(self):
with self.test_session():
x = tf.random_uniform((2, 3), seed=1)
kernel = partial(utils.gaussian_kernel_matrix,
sigmas=tf.constant([1.]))
loss = losses.maximum_mean_discrepancy(x, x, kernel)
self.assertEquals(loss.op.name, 'MaximumMeanDiscrepancy/value')
def test_mmd_is_zero_when_distributions_are_same(self):
with self.test_session():
x = tf.random_uniform((1000, 10), seed=1)
y = tf.random_uniform((1000, 10), seed=3)
kernel = partial(utils.gaussian_kernel_matrix, sigmas=tf.constant([100.]))
loss = losses.maximum_mean_discrepancy(x, y, kernel=kernel).eval()
self.assertAlmostEqual(0, loss, delta=1e-4)
def test_fast_mmd_is_similar_to_slow_mmd(self):
with self.test_session():
x = tf.constant(np.random.normal(size=(2, 3)), tf.float32)
y = tf.constant(np.random.rand(2, 3), tf.float32)
cost_old = MaximumMeanDiscrepancySlow(x, y, [1.]).eval()
kernel = partial(utils.gaussian_kernel_matrix, sigmas=tf.constant([1.]))
cost_new = losses.maximum_mean_discrepancy(x, y, kernel).eval()
self.assertAlmostEqual(cost_old, cost_new, delta=1e-5)
def test_fast_mmd_is_similar_to_slow_mmd(self):
with self.test_session():
x = tf.constant(np.random.normal(size=(2, 3)), tf.float32)
y = tf.constant(np.random.rand(2, 3), tf.float32)
cost_old = MaximumMeanDiscrepancySlow(x, y, [1.]).eval()
kernel = partial(utils.gaussian_kernel_matrix,
sigmas=tf.constant([1.]))
cost_new = losses.maximum_mean_discrepancy(x, y, kernel).eval()
self.assertAlmostEqual(cost_old, cost_new, delta=1e-5)
def test_mmd_is_zero_when_distributions_are_same(self):
with self.test_session():
x = tf.random_uniform((1000, 10), seed=1)
y = tf.random_uniform((1000, 10), seed=3)
kernel = partial(utils.gaussian_kernel_matrix,
sigmas=tf.constant([100.]))
loss = losses.maximum_mean_discrepancy(x, y, kernel=kernel).eval()
self.assertAlmostEqual(0, loss, delta=1e-4)
def test_multiple_sigmas(self):
with self.test_session():
x = tf.constant(np.random.normal(size=(2, 3)), tf.float32)
y = tf.constant(np.random.rand(2, 3), tf.float32)
sigmas = tf.constant([2., 5., 10, 20, 30])
kernel = partial(utils.gaussian_kernel_matrix, sigmas=sigmas)
cost_old = MaximumMeanDiscrepancySlow(x, y, [2., 5., 10, 20, 30]).eval()
cost_new = losses.maximum_mean_discrepancy(x, y, kernel=kernel).eval()
self.assertAlmostEqual(cost_old, cost_new, delta=1e-5)
def test_multiple_sigmas(self):
with self.test_session():
x = tf.constant(np.random.normal(size=(2, 3)), tf.float32)
y = tf.constant(np.random.rand(2, 3), tf.float32)
sigmas = tf.constant([2., 5., 10, 20, 30])
kernel = partial(utils.gaussian_kernel_matrix, sigmas=sigmas)
cost_old = MaximumMeanDiscrepancySlow(x, y,
[2., 5., 10, 20, 30]).eval()
cost_new = losses.maximum_mean_discrepancy(x, y,
kernel=kernel).eval()
self.assertAlmostEqual(cost_old, cost_new, delta=1e-5)
def __init_parameters(self):
self.tf_x = tf.placeholder(tf.float32, shape=[None, self.input_dim])
self.tf_source_data = tf.placeholder(tf.float32, shape=[None, self.input_dim])
self.tf_y = tf.placeholder(tf.float32, shape=[None, self.output_dim])
self.tf_z = tf.placeholder(tf.float32, shape=[None, self.latent_dim])
self.trainable_z = tf.get_variable('trainable_latent_value', shape=[self.query_num, self.latent_dim],
initializer=tf.random_normal_initializer(mean=0,stddev=1))
self.distribute_distance = maximum_mean_discrepancy(self.tf_x, self.tf_source_data)
self.GAN_params = self.gan(self.tf_x, self.tf_z)
trainable_fake_data = self.gan.generator(self.trainable_z, reuse=True)
self.Classifier_params = self.classifier(self.tf_x, self.tf_y, trainable_fake_data, \
self.trainable_z, self.tf_source_data)
self.sess = tf.Session()
self.saver = tf.train.Saver()
self.sess.run(tf.global_variables_initializer())
def test_mmd_is_zero_when_inputs_are_same(self):
with self.test_session():
x = tf.random_uniform((2, 3), seed=1)
kernel = partial(utils.gaussian_kernel_matrix, sigmas=tf.constant([1.]))
self.assertEquals(0, losses.maximum_mean_discrepancy(x, x, kernel).eval())