diff --git a/tests/ut/__init__.py b/tests/ut/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..1fec660695e211c236f0e3d252b54730c6e8f53b
--- /dev/null
+++ b/tests/ut/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for developed features of MindArmour.
+"""
diff --git a/tests/ut/python/__init__.py b/tests/ut/python/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..190835ec18d96820a9bd8702bfa2a02dc915ffa0
--- /dev/null
+++ b/tests/ut/python/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes python unit tests for developed features of MindArmour.
+"""
diff --git a/tests/ut/python/adv_robustness/__init__.py b/tests/ut/python/adv_robustness/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..02fa417b377dbf3995930ee95014e8e1673ed80a
--- /dev/null
+++ b/tests/ut/python/adv_robustness/__init__.py
@@ -0,0 +1,17 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for model attacks, model defenses and
+their evaluation functions.
+"""
diff --git a/tests/ut/python/adv_robustness/attacks/__init__.py b/tests/ut/python/adv_robustness/attacks/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..7c32a3086a2b61777b507dc67333b27556a514ad
--- /dev/null
+++ b/tests/ut/python/adv_robustness/attacks/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for white-box attack algorithms.
+"""
diff --git a/tests/ut/python/adv_robustness/attacks/black/__init__.py b/tests/ut/python/adv_robustness/attacks/black/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..a0256d6c0727fb3ba90d3cb934946ceaf7623279
--- /dev/null
+++ b/tests/ut/python/adv_robustness/attacks/black/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for black-box attack algorithms.
+"""
diff --git a/tests/ut/python/attacks/black/test_genetic_attack.py b/tests/ut/python/adv_robustness/attacks/black/test_genetic_attack.py
similarity index 100%
rename from tests/ut/python/attacks/black/test_genetic_attack.py
rename to tests/ut/python/adv_robustness/attacks/black/test_genetic_attack.py
diff --git a/tests/ut/python/attacks/black/test_hsja.py b/tests/ut/python/adv_robustness/attacks/black/test_hsja.py
similarity index 90%
rename from tests/ut/python/attacks/black/test_hsja.py
rename to tests/ut/python/adv_robustness/attacks/black/test_hsja.py
index b19e2032667dcc00ab91ecdcd879fe471717b913..d4ddaee277673df6b9966d46ceb145fb0df426e0 100644
--- a/tests/ut/python/attacks/black/test_hsja.py
+++ b/tests/ut/python/adv_robustness/attacks/black/test_hsja.py
@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import os
-import sys
 
 import numpy as np
 import pytest
@@ -24,9 +23,7 @@ from mindarmour import BlackModel
 from mindarmour.adv_robustness.attacks import HopSkipJumpAttack
 from mindarmour.utils.logger import LogUtil
 
-sys.path.append(os.path.join(os.path.dirname(os.path.abspath(__file__)),
-                             "../../../../../"))
-from example.mnist_demo.lenet5_net import LeNet5
+from ut.python.utils.mock_net import Net
 
 context.set_context(mode=context.GRAPH_MODE)
 context.set_context(device_target="Ascend")
@@ -64,25 +61,27 @@ def random_target_labels(true_labels):
 def create_target_images(dataset, data_labels, target_labels):
     res = []
     for label in target_labels:
-        for i in range(len(data_labels)):
-            if data_labels[i] == label:
+        for i, data_label in enumerate(data_labels):
+            if data_label == label:
                 res.append(dataset[i])
                 break
     return np.array(res)
 
+
 # public variable
 def get_model():
     # upload trained network
     current_dir = os.path.dirname(os.path.abspath(__file__))
     ckpt_name = os.path.join(current_dir,
-                             '../../test_data/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt')
-    net = LeNet5()
+                             '../../../dataset/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt')
+    net = Net()
     load_dict = load_checkpoint(ckpt_name)
     load_param_into_net(net, load_dict)
     net.set_train(False)
     model = ModelToBeAttacked(net)
     return model
 
+
 @pytest.mark.level0
 @pytest.mark.platform_arm_ascend_training
 @pytest.mark.platform_x86_ascend_training
@@ -97,9 +96,9 @@ def test_hsja_mnist_attack():
 
     # get test data
     test_images_set = np.load(os.path.join(current_dir,
-                                           '../../test_data/test_images.npy'))
+                                           '../../../dataset/test_images.npy'))
     test_labels_set = np.load(os.path.join(current_dir,
-                                           '../../test_data/test_labels.npy'))
+                                           '../../../dataset/test_labels.npy'))
     # prediction accuracy before attack
     model = get_model()
     batch_num = 1  # the number of batches of attacking samples
diff --git a/tests/ut/python/attacks/black/test_nes.py b/tests/ut/python/adv_robustness/attacks/black/test_nes.py
similarity index 92%
rename from tests/ut/python/attacks/black/test_nes.py
rename to tests/ut/python/adv_robustness/attacks/black/test_nes.py
index 6a849209d859224dbc04befa96507aaf760474f0..d9fa516a6e20bba5492763332490b7424111b86b 100644
--- a/tests/ut/python/attacks/black/test_nes.py
+++ b/tests/ut/python/adv_robustness/attacks/black/test_nes.py
@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import os
-import sys
 
 import numpy as np
 import pytest
@@ -24,9 +23,7 @@ from mindarmour import BlackModel
 from mindarmour.adv_robustness.attacks import NES
 from mindarmour.utils.logger import LogUtil
 
-sys.path.append(os.path.join(os.path.dirname(os.path.abspath(__file__)),
-                             "../../../../../"))
-from example.mnist_demo.lenet5_net import LeNet5
+from ut.python.utils.mock_net import Net
 
 context.set_context(mode=context.GRAPH_MODE)
 context.set_context(device_target="Ascend")
@@ -73,31 +70,35 @@ def _pseudorandom_target(index, total_indices, true_class):
 def create_target_images(dataset, data_labels, target_labels):
     res = []
     for label in target_labels:
-        for i in range(len(data_labels)):
-            if data_labels[i] == label:
+        for i, data_label in enumerate(data_labels):
+            if data_label == label:
                 res.append(dataset[i])
                 break
     return np.array(res)
 
+
 def get_model(current_dir):
     ckpt_name = os.path.join(current_dir,
-                             '../../test_data/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt')
-    net = LeNet5()
+                             '../../../dataset/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt')
+    net = Net()
     load_dict = load_checkpoint(ckpt_name)
     load_param_into_net(net, load_dict)
     net.set_train(False)
     model = ModelToBeAttacked(net)
     return model
 
+
 def get_dataset(current_dir):
     # upload trained network
 
     # get test data
     test_images = np.load(os.path.join(current_dir,
-                                       '../../test_data/test_images.npy'))
+                                       '../../../dataset/test_images.npy'))
     test_labels = np.load(os.path.join(current_dir,
-                                       '../../test_data/test_labels.npy'))
+                                       '../../../dataset/test_labels.npy'))
     return test_images, test_labels
+
+
 def nes_mnist_attack(scene, top_k):
     """
     hsja-Attack test
diff --git a/tests/ut/python/attacks/black/test_pointwise_attack.py b/tests/ut/python/adv_robustness/attacks/black/test_pointwise_attack.py
similarity index 86%
rename from tests/ut/python/attacks/black/test_pointwise_attack.py
rename to tests/ut/python/adv_robustness/attacks/black/test_pointwise_attack.py
index 28c2c061add1161efc4f250fecc77f43cd00b580..6428cdb144b8e2122debb2fab1ad3a50800d895c 100644
--- a/tests/ut/python/attacks/black/test_pointwise_attack.py
+++ b/tests/ut/python/adv_robustness/attacks/black/test_pointwise_attack.py
@@ -15,7 +15,6 @@
 PointWise Attack test
 """
 import os
-import sys
 
 import numpy as np
 import pytest
@@ -27,10 +26,7 @@ from mindarmour import BlackModel
 from mindarmour.adv_robustness.attacks import PointWiseAttack
 from mindarmour.utils.logger import LogUtil
 
-sys.path.append(os.path.join(os.path.dirname(os.path.abspath(__file__)),
-                             "../../../../../"))
-from example.mnist_demo.lenet5_net import LeNet5
-
+from ut.python.utils.mock_net import Net
 
 context.set_context(mode=context.GRAPH_MODE, device_target="Ascend")
 
@@ -65,16 +61,16 @@ def test_pointwise_attack_method():
     # upload trained network
     current_dir = os.path.dirname(os.path.abspath(__file__))
     ckpt_name = os.path.join(current_dir,
-                             '../../test_data/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt')
-    net = LeNet5()
+                             '../../../dataset/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt')
+    net = Net()
     load_dict = load_checkpoint(ckpt_name)
     load_param_into_net(net, load_dict)
 
     # get one mnist image
     input_np = np.load(os.path.join(current_dir,
-                                    '../../test_data/test_images.npy'))[:3]
+                                    '../../../dataset/test_images.npy'))[:3]
     labels = np.load(os.path.join(current_dir,
-                                  '../../test_data/test_labels.npy'))[:3]
+                                  '../../../dataset/test_labels.npy'))[:3]
     model = ModelToBeAttacked(net)
     pre_label = np.argmax(model.predict(input_np), axis=1)
     LOGGER.info(TAG, 'original sample predict labels are :{}'.format(pre_label))
diff --git a/tests/ut/python/attacks/black/test_pso_attack.py b/tests/ut/python/adv_robustness/attacks/black/test_pso_attack.py
similarity index 100%
rename from tests/ut/python/attacks/black/test_pso_attack.py
rename to tests/ut/python/adv_robustness/attacks/black/test_pso_attack.py
diff --git a/tests/ut/python/attacks/black/test_salt_and_pepper_attack.py b/tests/ut/python/adv_robustness/attacks/black/test_salt_and_pepper_attack.py
similarity index 98%
rename from tests/ut/python/attacks/black/test_salt_and_pepper_attack.py
rename to tests/ut/python/adv_robustness/attacks/black/test_salt_and_pepper_attack.py
index ad08b1b6ad537b2e0c093c4fee77bde831dfece8..28b4743c5ba9ee1f720079d0d7623a741a9999cc 100644
--- a/tests/ut/python/attacks/black/test_salt_and_pepper_attack.py
+++ b/tests/ut/python/adv_robustness/attacks/black/test_salt_and_pepper_attack.py
@@ -90,7 +90,7 @@ def test_salt_and_pepper_attack_method():
     labels = labels.astype(np.float32)
 
     attack = SaltAndPepperNoiseAttack(model, sparse=False)
-    is_adv, adv_data, query_times = attack.generate(inputs, labels)
+    _, adv_data, _ = attack.generate(inputs, labels)
     assert np.any(adv_data[0] != inputs[0]), 'Salt and pepper attack method: ' \
                                              'generate value must not be equal' \
                                              ' to original value.'
diff --git a/tests/ut/python/attacks/test_batch_generate_attack.py b/tests/ut/python/adv_robustness/attacks/test_batch_generate_attack.py
similarity index 100%
rename from tests/ut/python/attacks/test_batch_generate_attack.py
rename to tests/ut/python/adv_robustness/attacks/test_batch_generate_attack.py
diff --git a/tests/ut/python/attacks/test_cw.py b/tests/ut/python/adv_robustness/attacks/test_cw.py
similarity index 100%
rename from tests/ut/python/attacks/test_cw.py
rename to tests/ut/python/adv_robustness/attacks/test_cw.py
diff --git a/tests/ut/python/attacks/test_deep_fool.py b/tests/ut/python/adv_robustness/attacks/test_deep_fool.py
similarity index 100%
rename from tests/ut/python/attacks/test_deep_fool.py
rename to tests/ut/python/adv_robustness/attacks/test_deep_fool.py
diff --git a/tests/ut/python/attacks/test_gradient_method.py b/tests/ut/python/adv_robustness/attacks/test_gradient_method.py
similarity index 100%
rename from tests/ut/python/attacks/test_gradient_method.py
rename to tests/ut/python/adv_robustness/attacks/test_gradient_method.py
diff --git a/tests/ut/python/attacks/test_iterative_gradient_method.py b/tests/ut/python/adv_robustness/attacks/test_iterative_gradient_method.py
similarity index 100%
rename from tests/ut/python/attacks/test_iterative_gradient_method.py
rename to tests/ut/python/adv_robustness/attacks/test_iterative_gradient_method.py
diff --git a/tests/ut/python/attacks/test_jsma.py b/tests/ut/python/adv_robustness/attacks/test_jsma.py
similarity index 100%
rename from tests/ut/python/attacks/test_jsma.py
rename to tests/ut/python/adv_robustness/attacks/test_jsma.py
diff --git a/tests/ut/python/attacks/test_lbfgs.py b/tests/ut/python/adv_robustness/attacks/test_lbfgs.py
similarity index 82%
rename from tests/ut/python/attacks/test_lbfgs.py
rename to tests/ut/python/adv_robustness/attacks/test_lbfgs.py
index 097df4e014f5bcc92d4f5d191d131b02b25ed5f0..46e7f0551852c82446dcb432cca1202507802e62 100644
--- a/tests/ut/python/attacks/test_lbfgs.py
+++ b/tests/ut/python/adv_robustness/attacks/test_lbfgs.py
@@ -15,7 +15,6 @@
 LBFGS-Attack test.
 """
 import os
-import sys
 
 import numpy as np
 import pytest
@@ -25,9 +24,7 @@ from mindspore.train.serialization import load_checkpoint, load_param_into_net
 from mindarmour.adv_robustness.attacks import LBFGS
 from mindarmour.utils.logger import LogUtil
 
-sys.path.append(os.path.join(os.path.dirname(os.path.abspath(__file__)),
-                             "../../../../"))
-from example.mnist_demo.lenet5_net import LeNet5
+from ut.python.utils.mock_net import Net
 
 context.set_context(mode=context.GRAPH_MODE, device_target="Ascend")
 
@@ -50,16 +47,16 @@ def test_lbfgs_attack():
     # upload trained network
     current_dir = os.path.dirname(os.path.abspath(__file__))
     ckpt_name = os.path.join(current_dir,
-                             '../test_data/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt')
-    net = LeNet5()
+                             '../../dataset/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt')
+    net = Net()
     load_dict = load_checkpoint(ckpt_name)
     load_param_into_net(net, load_dict)
 
     # get one mnist image
     input_np = np.load(os.path.join(current_dir,
-                                    '../test_data/test_images.npy'))[:1]
+                                    '../../dataset/test_images.npy'))[:1]
     label_np = np.load(os.path.join(current_dir,
-                                    '../test_data/test_labels.npy'))[:1]
+                                    '../../dataset/test_labels.npy'))[:1]
     LOGGER.debug(TAG, 'true label is :{}'.format(label_np[0]))
     classes = 10
     target_np = np.random.randint(0, classes, 1)
diff --git a/tests/ut/python/adv_robustness/defenses/__init__.py b/tests/ut/python/adv_robustness/defenses/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..93f240d2b8e51659ada297f9366cc219a6c96084
--- /dev/null
+++ b/tests/ut/python/adv_robustness/defenses/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for defense algorithms.
+"""
diff --git a/tests/ut/python/defenses/test_ad.py b/tests/ut/python/adv_robustness/defenses/test_ad.py
similarity index 98%
rename from tests/ut/python/defenses/test_ad.py
rename to tests/ut/python/adv_robustness/defenses/test_ad.py
index d13b809965f914671157472694792da1e53267cf..2c0470df2f89945e3a7c3a1bdb541694caccb811 100644
--- a/tests/ut/python/defenses/test_ad.py
+++ b/tests/ut/python/adv_robustness/defenses/test_ad.py
@@ -23,10 +23,11 @@ from mindspore import context
 from mindspore import nn
 from mindspore.nn.optim.momentum import Momentum
 
-from mock_net import Net
 from mindarmour.adv_robustness.defenses import AdversarialDefense
 from mindarmour.utils.logger import LogUtil
 
+from ut.python.utils.mock_net import Net
+
 LOGGER = LogUtil.get_instance()
 TAG = 'Ad_Test'
 
diff --git a/tests/ut/python/defenses/test_ead.py b/tests/ut/python/adv_robustness/defenses/test_ead.py
similarity index 98%
rename from tests/ut/python/defenses/test_ead.py
rename to tests/ut/python/adv_robustness/defenses/test_ead.py
index ecf9dfa6875cf7cffa281de4cca24c051475a437..eb057e79d7fd8af1c072e93871e51352f9ae98e7 100644
--- a/tests/ut/python/defenses/test_ead.py
+++ b/tests/ut/python/adv_robustness/defenses/test_ead.py
@@ -22,13 +22,14 @@ from mindspore import context
 from mindspore import nn
 from mindspore.nn.optim.momentum import Momentum
 
-from mock_net import Net
 from mindarmour.adv_robustness.attacks import FastGradientSignMethod
 from mindarmour.adv_robustness.attacks import \
     ProjectedGradientDescent
 from mindarmour.adv_robustness.defenses import EnsembleAdversarialDefense
 from mindarmour.utils.logger import LogUtil
 
+from ut.python.utils.mock_net import Net
+
 LOGGER = LogUtil.get_instance()
 TAG = 'Ead_Test'
 
diff --git a/tests/ut/python/defenses/test_nad.py b/tests/ut/python/adv_robustness/defenses/test_nad.py
similarity index 97%
rename from tests/ut/python/defenses/test_nad.py
rename to tests/ut/python/adv_robustness/defenses/test_nad.py
index c61fef8cf707a059341363a0b84fecef12994e5a..8c0e5e030deeb374de1061cb6e891f81c912e7ec 100644
--- a/tests/ut/python/defenses/test_nad.py
+++ b/tests/ut/python/adv_robustness/defenses/test_nad.py
@@ -22,10 +22,11 @@ from mindspore import context
 from mindspore import nn
 from mindspore.nn.optim.momentum import Momentum
 
-from mock_net import Net
 from mindarmour.adv_robustness.defenses import NaturalAdversarialDefense
 from mindarmour.utils.logger import LogUtil
 
+from ut.python.utils.mock_net import Net
+
 LOGGER = LogUtil.get_instance()
 TAG = 'Nad_Test'
 
diff --git a/tests/ut/python/defenses/test_pad.py b/tests/ut/python/adv_robustness/defenses/test_pad.py
similarity index 98%
rename from tests/ut/python/defenses/test_pad.py
rename to tests/ut/python/adv_robustness/defenses/test_pad.py
index c8ea77d77d519ab594f960428aa25d469ff4edee..fb4ed3a388d03522d8c589cfc47760c50c6f37b4 100644
--- a/tests/ut/python/defenses/test_pad.py
+++ b/tests/ut/python/adv_robustness/defenses/test_pad.py
@@ -22,10 +22,11 @@ from mindspore import context
 from mindspore import nn
 from mindspore.nn.optim.momentum import Momentum
 
-from mock_net import Net
 from mindarmour.adv_robustness.defenses import ProjectedAdversarialDefense
 from mindarmour.utils.logger import LogUtil
 
+from ut.python.utils.mock_net import Net
+
 LOGGER = LogUtil.get_instance()
 TAG = 'Pad_Test'
 
diff --git a/tests/ut/python/adv_robustness/detectors/__init__.py b/tests/ut/python/adv_robustness/detectors/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..2cecd48e7942c8da6326a8de0d2627d68eea2e79
--- /dev/null
+++ b/tests/ut/python/adv_robustness/detectors/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for detection algorithms of white-box attacks.
+"""
diff --git a/tests/ut/python/adv_robustness/detectors/black/__init__.py b/tests/ut/python/adv_robustness/detectors/black/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..443a34a4eed4bf773776bca3ee3abd0749eed672
--- /dev/null
+++ b/tests/ut/python/adv_robustness/detectors/black/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for detection algorithms of black-box attacks.
+"""
diff --git a/tests/ut/python/detectors/black/test_similarity_detector.py b/tests/ut/python/adv_robustness/detectors/black/test_similarity_detector.py
similarity index 100%
rename from tests/ut/python/detectors/black/test_similarity_detector.py
rename to tests/ut/python/adv_robustness/detectors/black/test_similarity_detector.py
diff --git a/tests/ut/python/detectors/test_ensemble_detector.py b/tests/ut/python/adv_robustness/detectors/test_ensemble_detector.py
similarity index 100%
rename from tests/ut/python/detectors/test_ensemble_detector.py
rename to tests/ut/python/adv_robustness/detectors/test_ensemble_detector.py
diff --git a/tests/ut/python/detectors/test_mag_net.py b/tests/ut/python/adv_robustness/detectors/test_mag_net.py
similarity index 100%
rename from tests/ut/python/detectors/test_mag_net.py
rename to tests/ut/python/adv_robustness/detectors/test_mag_net.py
diff --git a/tests/ut/python/detectors/test_region_based_detector.py b/tests/ut/python/adv_robustness/detectors/test_region_based_detector.py
similarity index 100%
rename from tests/ut/python/detectors/test_region_based_detector.py
rename to tests/ut/python/adv_robustness/detectors/test_region_based_detector.py
diff --git a/tests/ut/python/detectors/test_spatial_smoothing.py b/tests/ut/python/adv_robustness/detectors/test_spatial_smoothing.py
similarity index 100%
rename from tests/ut/python/detectors/test_spatial_smoothing.py
rename to tests/ut/python/adv_robustness/detectors/test_spatial_smoothing.py
diff --git a/tests/ut/python/adv_robustness/evaluations/__init__.py b/tests/ut/python/adv_robustness/evaluations/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..c8b4b13652f7de306dd44cbf50c0a8ecf8e59adf
--- /dev/null
+++ b/tests/ut/python/adv_robustness/evaluations/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for the evaluations of white-box attack and defense.
+"""
diff --git a/tests/ut/python/adv_robustness/evaluations/black/__init__.py b/tests/ut/python/adv_robustness/evaluations/black/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..4f54ce8d9e69a14a966b51f4f7f19c446b8fd9b3
--- /dev/null
+++ b/tests/ut/python/adv_robustness/evaluations/black/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for the evaluations of black-box attack and defense.
+"""
diff --git a/tests/ut/python/evaluations/black/test_black_defense_eval.py b/tests/ut/python/adv_robustness/evaluations/black/test_black_defense_eval.py
similarity index 100%
rename from tests/ut/python/evaluations/black/test_black_defense_eval.py
rename to tests/ut/python/adv_robustness/evaluations/black/test_black_defense_eval.py
diff --git a/tests/ut/python/evaluations/test_attack_eval.py b/tests/ut/python/adv_robustness/evaluations/test_attack_eval.py
similarity index 100%
rename from tests/ut/python/evaluations/test_attack_eval.py
rename to tests/ut/python/adv_robustness/evaluations/test_attack_eval.py
diff --git a/tests/ut/python/evaluations/test_defense_eval.py b/tests/ut/python/adv_robustness/evaluations/test_defense_eval.py
similarity index 100%
rename from tests/ut/python/evaluations/test_defense_eval.py
rename to tests/ut/python/adv_robustness/evaluations/test_defense_eval.py
diff --git a/tests/ut/python/evaluations/test_radar_metric.py b/tests/ut/python/adv_robustness/evaluations/test_radar_metric.py
similarity index 98%
rename from tests/ut/python/evaluations/test_radar_metric.py
rename to tests/ut/python/adv_robustness/evaluations/test_radar_metric.py
index f43db5ec89a2678afe3a259c6303f5286f15a95e..bdcb12058805da18f845bc81fb1f1667ad66216e 100644
--- a/tests/ut/python/evaluations/test_radar_metric.py
+++ b/tests/ut/python/adv_robustness/evaluations/test_radar_metric.py
@@ -16,6 +16,8 @@ Radar map test.
 """
 import pytest
 from mindarmour.adv_robustness.evaluations import RadarMetric
+
+
 @pytest.mark.level0
 @pytest.mark.platform_arm_ascend_training
 @pytest.mark.platform_x86_ascend_training
@@ -31,7 +33,7 @@ def test_radar_metric():
 
     # create obj
     _ = RadarMetric(metrics_name, metrics_data, metrics_labels, title='',
-                     scale='sparse')
+                    scale='sparse')
 
 
 @pytest.mark.level0
diff --git a/tests/ut/python/test_data/test_images.npy b/tests/ut/python/dataset/test_images.npy
similarity index 100%
rename from tests/ut/python/test_data/test_images.npy
rename to tests/ut/python/dataset/test_images.npy
diff --git a/tests/ut/python/test_data/test_labels.npy b/tests/ut/python/dataset/test_labels.npy
similarity index 100%
rename from tests/ut/python/test_data/test_labels.npy
rename to tests/ut/python/dataset/test_labels.npy
diff --git a/tests/ut/python/test_data/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt b/tests/ut/python/dataset/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt
similarity index 100%
rename from tests/ut/python/test_data/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt
rename to tests/ut/python/dataset/trained_ckpt_file/checkpoint_lenet-10_1875.ckpt
diff --git a/tests/ut/python/defenses/mock_net.py b/tests/ut/python/defenses/mock_net.py
deleted file mode 100644
index db49ebc2fc66e73783ee6b91694935f27fda94a5..0000000000000000000000000000000000000000
--- a/tests/ut/python/defenses/mock_net.py
+++ /dev/null
@@ -1,105 +0,0 @@
-# Copyright 2019 Huawei Technologies Co., Ltd
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-# http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-"""
-mocked model for UT of defense algorithms.
-"""
-import numpy as np
-
-from mindspore import nn
-from mindspore import Tensor
-from mindspore.nn import WithLossCell, TrainOneStepCell
-from mindspore.nn.optim.momentum import Momentum
-from mindspore import context
-from mindspore.common.initializer import TruncatedNormal
-
-from mindarmour.adv_robustness.attacks import FastGradientSignMethod
-
-
-def conv(in_channels, out_channels, kernel_size, stride=1, padding=0):
-    weight = weight_variable()
-    return nn.Conv2d(in_channels, out_channels,
-                     kernel_size=kernel_size, stride=stride, padding=padding,
-                     weight_init=weight, has_bias=False, pad_mode="valid")
-
-
-def fc_with_initialize(input_channels, out_channels):
-    weight = weight_variable()
-    bias = weight_variable()
-    return nn.Dense(input_channels, out_channels, weight, bias)
-
-
-def weight_variable():
-    return TruncatedNormal(0.02)
-
-
-class Net(nn.Cell):
-    """
-    Lenet network
-    """
-    def __init__(self):
-        super(Net, self).__init__()
-        self.conv1 = conv(1, 6, 5)
-        self.conv2 = conv(6, 16, 5)
-        self.fc1 = fc_with_initialize(16*5*5, 120)
-        self.fc2 = fc_with_initialize(120, 84)
-        self.fc3 = fc_with_initialize(84, 10)
-        self.relu = nn.ReLU()
-        self.max_pool2d = nn.MaxPool2d(kernel_size=2, stride=2)
-        self.flatten = nn.Flatten()
-
-    def construct(self, x):
-        x = self.conv1(x)
-        x = self.relu(x)
-        x = self.max_pool2d(x)
-        x = self.conv2(x)
-        x = self.relu(x)
-        x = self.max_pool2d(x)
-        x = self.flatten(x)
-        x = self.fc1(x)
-        x = self.relu(x)
-        x = self.fc2(x)
-        x = self.relu(x)
-        x = self.fc3(x)
-        return x
-
-
-if __name__ == '__main__':
-    num_classes = 10
-    batch_size = 32
-
-    sparse = False
-    context.set_context(mode=context.GRAPH_MODE)
-    context.set_context(device_target='Ascend')
-
-    # create test data
-    inputs_np = np.random.rand(batch_size, 1, 32, 32).astype(np.float32)
-    labels_np = np.random.randint(num_classes, size=batch_size).astype(np.int32)
-    if not sparse:
-        labels_np = np.eye(num_classes)[labels_np].astype(np.float32)
-
-    net = Net()
-
-    # test fgsm
-    attack = FastGradientSignMethod(net, eps=0.3)
-    attack.generate(inputs_np, labels_np)
-
-    # test train ops
-    loss_fn = nn.SoftmaxCrossEntropyWithLogits(sparse=sparse)
-    optimizer = Momentum(filter(lambda x: x.requires_grad, net.get_parameters()),
-                         0.01, 0.9)
-    loss_net = WithLossCell(net, loss_fn)
-    train_net = TrainOneStepCell(loss_net, optimizer)
-    train_net.set_train()
-
-    train_net(Tensor(inputs_np), Tensor(labels_np))
diff --git a/tests/ut/python/fuzzing/__init__.py b/tests/ut/python/fuzzing/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..41fabd7256c40302777d1c90599fd31c955f821e
--- /dev/null
+++ b/tests/ut/python/fuzzing/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for model fuzzing.
+"""
diff --git a/tests/ut/python/privacy/__init__.py b/tests/ut/python/privacy/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..d57ca645472b2213ef74e4d407fcef7a0bb07cfc
--- /dev/null
+++ b/tests/ut/python/privacy/__init__.py
@@ -0,0 +1,17 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for differential-privacy training and
+privacy breach estimation.
+"""
diff --git a/tests/ut/python/privacy/diff_privacy/__init__.py b/tests/ut/python/privacy/diff_privacy/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..89d9484ed71c5e12aa5d3c7e9a1f7aea75efd724
--- /dev/null
+++ b/tests/ut/python/privacy/diff_privacy/__init__.py
@@ -0,0 +1,17 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for differential-privacy training and
+privacy-budget estimation.
+"""
diff --git a/tests/ut/python/diff_privacy/test_mechanisms.py b/tests/ut/python/privacy/diff_privacy/test_mechanisms.py
similarity index 100%
rename from tests/ut/python/diff_privacy/test_mechanisms.py
rename to tests/ut/python/privacy/diff_privacy/test_mechanisms.py
diff --git a/tests/ut/python/diff_privacy/test_model_train.py b/tests/ut/python/privacy/diff_privacy/test_model_train.py
similarity index 98%
rename from tests/ut/python/diff_privacy/test_model_train.py
rename to tests/ut/python/privacy/diff_privacy/test_model_train.py
index 7608aaa0d4d9e82293b739a7db1924d82c4006f2..6211adef54646bb439c4ea95a3da270ca68d7e3f 100644
--- a/tests/ut/python/diff_privacy/test_model_train.py
+++ b/tests/ut/python/privacy/diff_privacy/test_model_train.py
@@ -26,7 +26,7 @@ from mindarmour.privacy.diff_privacy import NoiseMechanismsFactory
 from mindarmour.privacy.diff_privacy import ClipMechanismsFactory
 from mindarmour.privacy.diff_privacy import DPOptimizerClassFactory
 
-from test_network import LeNet5
+from ut.python.utils.mock_net import Net
 
 
 def dataset_generator(batch_size, batches):
@@ -48,7 +48,7 @@ def test_dp_model_with_pynative_mode():
     context.set_context(mode=context.PYNATIVE_MODE, device_target="Ascend")
     norm_bound = 1.0
     initial_noise_multiplier = 0.01
-    network = LeNet5()
+    network = Net()
     batch_size = 32
     batches = 128
     epochs = 1
@@ -88,7 +88,7 @@ def test_dp_model_with_graph_mode():
     context.set_context(mode=context.GRAPH_MODE, device_target="Ascend")
     norm_bound = 1.0
     initial_noise_multiplier = 0.01
-    network = LeNet5()
+    network = Net()
     batch_size = 32
     batches = 128
     epochs = 1
@@ -126,7 +126,7 @@ def test_dp_model_with_graph_mode_ada_gaussian():
     context.set_context(mode=context.GRAPH_MODE, device_target="Ascend")
     norm_bound = 1.0
     initial_noise_multiplier = 0.01
-    network = LeNet5()
+    network = Net()
     batch_size = 32
     batches = 128
     epochs = 1
diff --git a/tests/ut/python/diff_privacy/test_monitor.py b/tests/ut/python/privacy/diff_privacy/test_monitor.py
similarity index 98%
rename from tests/ut/python/diff_privacy/test_monitor.py
rename to tests/ut/python/privacy/diff_privacy/test_monitor.py
index b0cc0cc9e6e5a58a5d8bfc77c08cc767aa4d67af..855796934fe8a75ef57aa9da2391b16175c6bff5 100644
--- a/tests/ut/python/diff_privacy/test_monitor.py
+++ b/tests/ut/python/privacy/diff_privacy/test_monitor.py
@@ -25,7 +25,7 @@ import mindspore.context as context
 from mindarmour.privacy.diff_privacy import PrivacyMonitorFactory
 from mindarmour.utils.logger import LogUtil
 
-from test_network import LeNet5
+from ut.python.utils.mock_net import Net
 
 LOGGER = LogUtil.get_instance()
 TAG = 'DP-Monitor Test'
@@ -57,7 +57,7 @@ def test_dp_monitor():
     suggest_epoch = rdp.max_epoch_suggest()
     LOGGER.info(TAG, 'The recommended maximum training epochs is: %s',
                 suggest_epoch)
-    network = LeNet5()
+    network = Net()
     net_loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction="mean")
     net_opt = nn.Momentum(network.trainable_params(), 0.01, 0.9)
 
@@ -86,7 +86,7 @@ def test_dp_monitor_gpu():
     suggest_epoch = rdp.max_epoch_suggest()
     LOGGER.info(TAG, 'The recommended maximum training epochs is: %s',
                 suggest_epoch)
-    network = LeNet5()
+    network = Net()
     net_loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction="mean")
     net_opt = nn.Momentum(network.trainable_params(), 0.01, 0.9)
 
@@ -115,7 +115,7 @@ def test_dp_monitor_cpu():
     suggest_epoch = rdp.max_epoch_suggest()
     LOGGER.info(TAG, 'The recommended maximum training epochs is: %s',
                 suggest_epoch)
-    network = LeNet5()
+    network = Net()
     net_loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction="mean")
     net_opt = nn.Momentum(network.trainable_params(), 0.01, 0.9)
 
@@ -145,7 +145,7 @@ def test_dp_monitor_zcdp():
     suggest_epoch = zcdp.max_epoch_suggest()
     LOGGER.info(TAG, 'The recommended maximum training epochs is: %s',
                 suggest_epoch)
-    network = LeNet5()
+    network = Net()
     net_loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction="mean")
     net_opt = nn.Momentum(network.trainable_params(), 0.01, 0.9)
 
@@ -174,7 +174,7 @@ def test_dp_monitor_zcdp_gpu():
     suggest_epoch = zcdp.max_epoch_suggest()
     LOGGER.info(TAG, 'The recommended maximum training epochs is: %s',
                 suggest_epoch)
-    network = LeNet5()
+    network = Net()
     net_loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction="mean")
     net_opt = nn.Momentum(network.trainable_params(), 0.01, 0.9)
 
@@ -203,7 +203,7 @@ def test_dp_monitor_zcdp_cpu():
     suggest_epoch = zcdp.max_epoch_suggest()
     LOGGER.info(TAG, 'The recommended maximum training epochs is: %s',
                 suggest_epoch)
-    network = LeNet5()
+    network = Net()
     net_loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction="mean")
     net_opt = nn.Momentum(network.trainable_params(), 0.01, 0.9)
 
diff --git a/tests/ut/python/diff_privacy/test_optimizer.py b/tests/ut/python/privacy/diff_privacy/test_optimizer.py
similarity index 96%
rename from tests/ut/python/diff_privacy/test_optimizer.py
rename to tests/ut/python/privacy/diff_privacy/test_optimizer.py
index 128ba0e2dd8337299c5152ddccc9243dfc218fb9..ca0831329d502fcccceaa6efe79afc2312d34124 100644
--- a/tests/ut/python/diff_privacy/test_optimizer.py
+++ b/tests/ut/python/privacy/diff_privacy/test_optimizer.py
@@ -19,7 +19,7 @@ from mindspore.train.model import Model
 
 from mindarmour.privacy.diff_privacy import DPOptimizerClassFactory
 
-from test_network import LeNet5
+from ut.python.utils.mock_net import Net
 
 
 @pytest.mark.level0
@@ -29,7 +29,7 @@ from test_network import LeNet5
 @pytest.mark.component_mindarmour
 def test_optimizer():
     context.set_context(mode=context.PYNATIVE_MODE, device_target="Ascend")
-    network = LeNet5()
+    network = Net()
     lr = 0.01
     momentum = 0.9
     micro_batches = 2
@@ -47,7 +47,7 @@ def test_optimizer():
 @pytest.mark.component_mindarmour
 def test_optimizer_gpu():
     context.set_context(mode=context.PYNATIVE_MODE, device_target="GPU")
-    network = LeNet5()
+    network = Net()
     lr = 0.01
     momentum = 0.9
     micro_batches = 2
@@ -65,7 +65,7 @@ def test_optimizer_gpu():
 @pytest.mark.component_mindarmour
 def test_optimizer_cpu():
     context.set_context(mode=context.PYNATIVE_MODE, device_target="CPU")
-    network = LeNet5()
+    network = Net()
     lr = 0.01
     momentum = 0.9
     micro_batches = 2
diff --git a/tests/ut/python/privacy/evaluation/__init__.py b/tests/ut/python/privacy/evaluation/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..dc7bb709a42c0558cde74125f5c389c97cb34160
--- /dev/null
+++ b/tests/ut/python/privacy/evaluation/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes unit tests for privacy breach estimation.
+"""
diff --git a/tests/ut/python/diff_privacy/test_attacker.py b/tests/ut/python/privacy/evaluation/test_attacker.py
similarity index 100%
rename from tests/ut/python/diff_privacy/test_attacker.py
rename to tests/ut/python/privacy/evaluation/test_attacker.py
diff --git a/tests/ut/python/diff_privacy/test_membership_inference.py b/tests/ut/python/privacy/evaluation/test_membership_inference.py
similarity index 96%
rename from tests/ut/python/diff_privacy/test_membership_inference.py
rename to tests/ut/python/privacy/evaluation/test_membership_inference.py
index d4f925c4e08dc372115318bd3ce1a7c365d97478..1065f84eef599a2ed406c83af1175c6ac6d628a1 100644
--- a/tests/ut/python/diff_privacy/test_membership_inference.py
+++ b/tests/ut/python/privacy/evaluation/test_membership_inference.py
@@ -14,9 +14,6 @@
 """
 membership inference test
 """
-import os
-import sys
-
 import pytest
 
 import numpy as np
@@ -28,12 +25,12 @@ import mindspore.context as context
 
 from mindarmour.privacy.evaluation import MembershipInference
 
-sys.path.append(os.path.join(os.path.dirname(os.path.abspath(__file__)), "../"))
-from defenses.mock_net import Net
+from ut.python.utils.mock_net import Net
 
 
 context.set_context(mode=context.GRAPH_MODE)
 
+
 def dataset_generator(batch_size, batches):
     """mock training data."""
     data = np.random.randn(batches*batch_size, 1, 32, 32).astype(
diff --git a/tests/ut/python/utils/__init__.py b/tests/ut/python/utils/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..ea65540163b90fb2e29ae2bf10dac35458b28479
--- /dev/null
+++ b/tests/ut/python/utils/__init__.py
@@ -0,0 +1,16 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This package includes common methods called by unit tests.
+"""
diff --git a/tests/ut/python/diff_privacy/test_network.py b/tests/ut/python/utils/mock_net.py
similarity index 93%
rename from tests/ut/python/diff_privacy/test_network.py
rename to tests/ut/python/utils/mock_net.py
index ac5cf3bd095feeda765edffa888385c6e2a2368a..7b5b4bb9564cabb70728a266fa6ba5d75d2a4dff 100644
--- a/tests/ut/python/diff_privacy/test_network.py
+++ b/tests/ut/python/utils/mock_net.py
@@ -11,6 +11,9 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
+"""
+mocked model for UT of defense algorithms.
+"""
 from mindspore import nn
 from mindspore.common.initializer import TruncatedNormal
 
@@ -29,15 +32,15 @@ def fc_with_initialize(input_channels, out_channels):
 
 
 def weight_variable():
-    return TruncatedNormal(0.05)
+    return TruncatedNormal(0.02)
 
 
-class LeNet5(nn.Cell):
+class Net(nn.Cell):
     """
     Lenet network
     """
     def __init__(self):
-        super(LeNet5, self).__init__()
+        super(Net, self).__init__()
         self.conv1 = conv(1, 6, 5)
         self.conv2 = conv(6, 16, 5)
         self.fc1 = fc_with_initialize(16*5*5, 120)