def test_value_error():
np.random.seed(5)
ori = np.random.rand(4, 4).astype(np.float32)
labels = np.array([[1, 0, 0, 0], [0, 0, 1, 0], [0, 0, 1, 0],
[0, 1, 0, 0]]).astype(np.int32)
np.random.seed(6)
adv = np.random.rand(4, 4).astype(np.float32)
model = Model(Net())
# model should be mindspore model
with pytest.raises(TypeError):
assert RegionBasedDetector(Net())
with pytest.raises(ValueError):
assert RegionBasedDetector(model, number_points=-1)
with pytest.raises(ValueError):
assert RegionBasedDetector(model, initial_radius=-1)
with pytest.raises(ValueError):
assert RegionBasedDetector(model, max_radius=-2.2)
with pytest.raises(ValueError):
assert RegionBasedDetector(model, search_step=0)
with pytest.raises(TypeError):
assert RegionBasedDetector(model, sparse='False')
detector = RegionBasedDetector(model)
with pytest.raises(TypeError):
# radius must not empty
assert detector.detect(adv)
radius = detector.fit(ori, labels)
detector.set_radius(radius)
with pytest.raises(TypeError):
# adv type should be in (list, tuple, numpy.ndarray)
assert detector.detect(adv.tostring())
def test_region_based_classification():
"""
Compute mindspore result.
"""
np.random.seed(5)
ori = np.random.rand(4, 4).astype(np.float32)
labels = np.array([[1, 0, 0, 0], [0, 0, 1, 0], [0, 0, 1, 0],
[0, 1, 0, 0]]).astype(np.int32)
np.random.seed(6)
adv = np.random.rand(4, 4).astype(np.float32)
model = Model(Net())
detector = RegionBasedDetector(model)
radius = detector.fit(ori, labels)
detector.set_radius(radius)
detected_res = detector.detect(adv)
expected_value = np.array([0, 0, 1, 0])
assert np.all(detected_res == expected_value)