def test_loss_augmentation():
X, Y = generate_blocks(n_samples=1)
x, y = X[0], Y[0]
w = np.array([1, 0, 0, 1, 0, -4, 0]) # unary # pairwise
crf = GridCRF()
crf.initialize(X, Y)
y_hat, energy = crf.loss_augmented_inference(x, y, w, return_energy=True)
assert_almost_equal(energy + crf.loss(y, y_hat), -np.dot(w, crf.joint_feature(x, y_hat)))
def test_loss_augmentation():
X, Y = toy.generate_blocks(n_samples=1)
x, y = X[0], Y[0]
w = np.array([1, 0, # unary
0, 1,
0, # pairwise
-4, 0])
crf = GridCRF(inference_method='lp')
y_hat, energy = crf.loss_augmented_inference(x, y, w, return_energy=True)
assert_almost_equal(energy + crf.loss(y, y_hat),
-np.dot(w, crf.psi(x, y_hat)))
def test_loss_augmentation():
X, Y = generate_blocks(n_samples=1)
x, y = X[0], Y[0]
w = np.array([1, 0, # unary
0, 1,
0, # pairwise
-4, 0])
crf = GridCRF()
crf.initialize(X, Y)
y_hat, energy = crf.loss_augmented_inference(x, y, w, return_energy=True)
assert_almost_equal(energy + crf.loss(y, y_hat),
-np.dot(w, crf.psi(x, y_hat)))
def test_binary_crf_exhaustive_loss_augmented():
# tests qpbo inference against brute force
# on random data / weights
np.random.seed(0)
for inference_method in get_installed(["qpbo", "lp"]):
crf = GridCRF(n_states=2, n_features=2, inference_method=inference_method)
for i in xrange(10):
# generate data and weights
y = np.random.randint(2, size=(3, 2))
x = np.random.uniform(-1, 1, size=(3, 2))
x = np.dstack([-x, np.zeros_like(x)])
w = np.random.uniform(-1, 1, size=7)
# check loss augmented map inference
y_hat = crf.loss_augmented_inference(x, y, w)
y_ex = exhaustive_loss_augmented_inference(crf, x, y, w)
assert_array_equal(y_hat, y_ex)
def test_binary_crf_exhaustive_loss_augmented():
# tests qpbo inference against brute force
# on random data / weights
np.random.seed(0)
for inference_method in get_installed(['qpbo', 'lp']):
crf = GridCRF(n_states=2, n_features=2,
inference_method=inference_method)
for i in xrange(10):
# generate data and weights
y = np.random.randint(2, size=(3, 2))
x = np.random.uniform(-1, 1, size=(3, 2))
x = np.dstack([-x, np.zeros_like(x)])
w = np.random.uniform(-1, 1, size=7)
# check loss augmented map inference
y_hat = crf.loss_augmented_inference(x, y, w)
y_ex = exhaustive_loss_augmented_inference(crf, x, y, w)
assert_array_equal(y_hat, y_ex)
def test_binary_crf_exhaustive_loss_augmented():
# tests graph cut inference against brute force
# on random data / weights
np.random.seed(0)
for inference_method in ['qpbo', 'lp']:
crf = GridCRF(inference_method=inference_method)
for i in xrange(50):
# generate data and weights
y = np.random.randint(2, size=(3, 3))
x = np.random.uniform(-1, 1, size=(3, 3))
x = np.dstack([-x, np.zeros_like(x)])
w = np.random.uniform(-1, 1, size=7)
# check loss augmented map inference
y_hat = crf.loss_augmented_inference(x, y, w)
y_ex = exhaustive_loss_augmented_inference(crf, x, y, w)
#print(y_hat)
#print(y_ex)
#print("++++++++++++++++++++++")
assert_array_equal(y_hat, y_ex)
