moved files

leakpro/tests/tests_mia/conftest.py

@@ -0,0 +1,69 @@
+"""Test for the image input handler."""
+import os
+import shutil
+
+import pytest
+import yaml
+from dotmap import DotMap
+
+from leakpro import LeakPro
+from leakpro.tests.constants import STORAGE_PATH, get_audit_config
+from leakpro.tests.input_handler.image_input_handler import ImageInputHandler
+from leakpro.tests.input_handler.image_utils import setup_image_test
+from leakpro.tests.input_handler.tabular_input_handler import TabularInputHandler
+from leakpro.tests.input_handler.tabular_utils import setup_tabular_test
+
+
+@pytest.fixture(scope="session")
+def manage_storage_directory():
+    """Fixture to create and remove the storage directory."""
+
+    # Setup: Create the folder at the start of the test session
+    os.makedirs(STORAGE_PATH, exist_ok=True)
+
+    # Yield control back to the test session
+    yield
+
+    # Teardown: Remove the folder and its contents at the end of the session
+    if os.path.exists(STORAGE_PATH):
+        shutil.rmtree(STORAGE_PATH)
+
+@pytest.fixture
+def image_handler(manage_storage_directory) -> ImageInputHandler:
+    """Fixture for the image input handler to be shared between many tests."""
+
+    config = DotMap()
+    config.target = setup_image_test()
+    config.audit = get_audit_config()
+    config.audit.modality = "image"
+    #save config to file
+    config_path = f"{STORAGE_PATH}/image_test_config.yaml"
+    with open(config_path, "w") as f:
+        yaml.dump(config.toDict(), f)
+
+    leakpro = LeakPro(ImageInputHandler, config_path)
+    handler = leakpro.handler
+    handler.configs = DotMap(handler.configs)
+
+    # Yield control back to the test session
+    return handler
+
+@pytest.fixture
+def tabular_handler(manage_storage_directory) -> TabularInputHandler:
+    """Fixture for the image input handler to be shared between many tests."""
+
+    config = DotMap()
+    config.target = setup_tabular_test()
+    config.audit = get_audit_config()
+    config.audit.modality = "tabular"
+    #save config to file
+    config_path = f"{STORAGE_PATH}/tabular_test_config.yaml"
+    with open(config_path, "w") as f:
+        yaml.dump(config.toDict(), f)
+
+    leakpro = LeakPro(TabularInputHandler, config_path)
+    handler = leakpro.handler
+    handler.configs = DotMap(handler.configs)
+
+    # Yield control back to the test session
+    return handler

leakpro/tests/tests_mia/constants.py

@@ -0,0 +1,65 @@
+from dotmap import DotMap
+
+STORAGE_PATH = "./leakpro/tests/tmp"
+
+# User input handler for images
+
+
+def get_image_handler_config():
+    parameters = DotMap()
+    parameters.target_folder = "./leakpro/tests/tmp/image"
+    parameters.epochs = 10
+    parameters.batch_size = 64
+    parameters.learning_rate = 0.001
+    parameters.optimizer = "sgd"
+    parameters.loss = "crossentropyloss"
+    parameters.data_points = 130
+    parameters.train_data_points = 20
+    parameters.test_data_points = 20
+    parameters.img_size = (3, 32, 32)
+    parameters.num_classes = 13
+    parameters.images_per_class = parameters.data_points // parameters.num_classes
+    return parameters
+
+def get_tabular_handler_config():
+    parameters = DotMap()
+    parameters.target_folder = "./leakpro/tests/tmp/tabular"
+    parameters.epochs = 10
+    parameters.batch_size = 64
+    parameters.learning_rate = 0.001
+    parameters.optimizer = "sgd"
+    parameters.loss = "BCEWithLogitsLoss"
+    parameters.data_points = 500
+    parameters.train_data_points = 200
+    parameters.test_data_points = 200
+    parameters.num_classes = 1
+    parameters.n_continuous = 10
+    parameters.n_categorical = 5
+    return parameters
+
+def get_audit_config():
+    #audit configuration
+    audit_config = DotMap()
+    audit_config.output_dir = STORAGE_PATH
+    audit_config.attack_type = "mia"
+    # Lira parameters
+    audit_config.attack_list.lira.training_data_fraction = 0.1
+    audit_config.attack_list.lira.num_shadow_models = 3
+    audit_config.attack_list.lira.online = False
+    audit_config.attack_list.lira.fixed_variance = True
+    return audit_config
+
+
+
+# Shadow model configuration for images
+def get_shadow_model_config():
+    shadow_model_config = DotMap()
+    shadow_model_config.module_path = "./leakpro/tests/input_handler/image_utils.py"
+    shadow_model_config.model_class = "ConvNet"
+    shadow_model_config.batch_size = 32
+    shadow_model_config.epochs = 1
+    shadow_model_config.optimizer = {"name": "sgd", "lr": 0.001}
+    shadow_model_config.loss = {"name": "crossentropyloss"}
+    return shadow_model_config
+
+