def test_local_pairwise_distances_two_dimensional(self):
a = np.array([[[1.0, -5.0], [7.0, 2.0]], [[1.0, 3.0], [3.0, 4.0]]])
b = np.array([[[2.0, 1.0], [0.0, -9.0]], [[4.0, 3.0], [3.0, 1.0]]])
g = tf.Graph()
with g.as_default():
with self.test_session(graph=g) as sess:
a_tf = tf.constant(a, dtype=tf.float32)
b_tf = tf.constant(b, dtype=tf.float32)
d = embedding_utils.local_pairwise_distances(a_tf,
b_tf,
max_distance=1)
d_val = sess.run(d)
for y in range(2):
for x in range(2):
for dy in range(-1, 2):
for dx in range(-1, 2):
a_slice = a[y, x, :]
if y + dy < 0 or y + dy > 1 or x + dx < 0 or x + dx > 1:
expected = np.float('inf')
else:
b_slice = b[y + dy, x + dx, :]
expected = ((a_slice - b_slice)**2).sum()
dy0 = dy + 1
dx0 = dx + 1
self.assertAlmostEqual(
d_val[y, x, 3 * dy0 + dx0], expected)
def test_local_pairwise_distances_shape(self):
a = np.zeros((4, 5, 2))
b = np.zeros((4, 5, 2))
g = tf.Graph()
with g.as_default():
with self.test_session(graph=g) as sess:
a_tf = tf.constant(a, dtype=tf.float32)
b_tf = tf.constant(b, dtype=tf.float32)
d = embedding_utils.local_pairwise_distances(a_tf, b_tf, max_distance=4)
d_val = sess.run(d)
self.assertAllEqual(d_val.shape, (4, 5, 81))
def test_local_pairwise_distances_shape(self):
a = np.zeros((4, 5, 2))
b = np.zeros((4, 5, 2))
g = tf.Graph()
with g.as_default():
with self.test_session(graph=g) as sess:
a_tf = tf.constant(a, dtype=tf.float32)
b_tf = tf.constant(b, dtype=tf.float32)
d = embedding_utils.local_pairwise_distances(a_tf,
b_tf,
max_distance=4)
d_val = sess.run(d)
self.assertAllEqual(d_val.shape, (4, 5, 81))
def test_local_pairwise_distances_two_dimensional(self):
a = np.array([[[1.0, -5.0], [7.0, 2.0]], [[1.0, 3.0], [3.0, 4.0]]])
b = np.array([[[2.0, 1.0], [0.0, -9.0]], [[4.0, 3.0], [3.0, 1.0]]])
g = tf.Graph()
with g.as_default():
with self.test_session(graph=g) as sess:
a_tf = tf.constant(a, dtype=tf.float32)
b_tf = tf.constant(b, dtype=tf.float32)
d = embedding_utils.local_pairwise_distances(a_tf, b_tf,
max_distance=1)
d_val = sess.run(d)
for y in range(2):
for x in range(2):
for dy in range(-1, 2):
for dx in range(-1, 2):
a_slice = a[y, x, :]
if y + dy < 0 or y + dy > 1 or x + dx < 0 or x + dx > 1:
expected = np.float('inf')
else:
b_slice = b[y + dy, x + dx, :]
expected = ((a_slice - b_slice) ** 2).sum()
dy0 = dy + 1
dx0 = dx + 1
self.assertAlmostEqual(d_val[y, x, 3 * dy0 + dx0], expected)