Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Improvements to attribute attack notebooks #2269

Merged
merged 1 commit into from
Sep 4, 2023
Merged

Improvements to attribute attack notebooks #2269

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 2 additions & 3 deletions notebooks/attack_attribute_inference.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,6 @@
"metadata": {},
"source": [
"## Preliminaries\n",
"In order to mount a successful attribute inference attack, the attacked feature must be categorical, and with a relatively small number of possible values (preferably binary, but should at least be less then the number of label classes).\n",
"\n",
"In the case of the nursery dataset, the sensitive feature we want to infer is the 'social' feature. In the original dataset this is a categorical feature with 3 possible values. To make the attack more successful, we reduced this to two possible feature values by assigning the original value 'problematic' the new value 1, and the other original values were assigned the new value 0.\n",
"\n",
Expand Down Expand Up @@ -391,7 +390,7 @@
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
Expand All @@ -405,7 +404,7 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.8.3"
"version": "3.9.6"
}
},
"nbformat": 4,
Expand Down
115 changes: 94 additions & 21 deletions notebooks/attack_attribute_inference_regressor.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,24 +4,22 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"# Running attribute inference attacks on Regression Models"
"# Running attribute inference attacks on regression models"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"In this tutorial we will show how to run black-box inference attacks on regression model. This will be demonstrated on the Nursery dataset (original dataset can be found here: https://archive.ics.uci.edu/ml/datasets/nursery). "
"In this tutorial we will show how to run a black-box attribute inference attack on a regression model. This will be demonstrated on the diabetes dataset from scikitlearn (https://scikit-learn.org/stable/datasets/toy_dataset.html#diabetes-dataset). "
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Preliminaries\n",
"In order to mount a successful attribute inference attack, the attacked feature must be categorical, and with a relatively small number of possible values (preferably binary).\n",
"\n",
"In the case of the diabetes dataset, the sensitive feature we want to infer is the 'sex' feature, which is a binary feature."
"## Attacking a categorical feature\n",
"We start by trying to infer the 'sex' feature, which is a binary feature."
]
},
{
Expand All @@ -33,14 +31,17 @@
},
{
"cell_type": "code",
"execution_count": 1,
"execution_count": 20,
"metadata": {},
"outputs": [],
"source": [
"import os\n",
"import sys\n",
"sys.path.insert(0, os.path.abspath('..'))\n",
"\n",
"import warnings\n",
"warnings.filterwarnings('ignore')\n",
"\n",
"from art.utils import load_diabetes\n",
"\n",
"(x_train, y_train), (x_test, y_test), _, _ = load_diabetes(test_set=0.5)"
Expand All @@ -55,14 +56,14 @@
},
{
"cell_type": "code",
"execution_count": 13,
"execution_count": 21,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Base model score: -0.04773984870966275\n"
"Base model score: -0.053305975661749994\n"
]
}
],
Expand Down Expand Up @@ -110,7 +111,7 @@
"# only attacked feature\n",
"attack_x_test_feature = attack_x_test[:, attack_feature].copy().reshape(-1, 1)\n",
"# training data without attacked feature\n",
"attack_x_test = np.delete(attack_x_test, attack_feature, 1)\n",
"x_test_for_attack = np.delete(attack_x_test, attack_feature, 1)\n",
"\n",
"bb_attack = AttributeInferenceBlackBox(art_regressor, attack_feature=attack_feature)\n",
"\n",
Expand All @@ -134,14 +135,14 @@
"name": "stdout",
"output_type": "stream",
"text": [
"0.5585585585585585\n"
"0.6126126126126126\n"
]
}
],
"source": [
"# get inferred values\n",
"values = [-0.88085106, 1.]\n",
"inferred_train_bb = bb_attack.infer(attack_x_test, pred=attack_x_test_predictions, values=values)\n",
"inferred_train_bb = bb_attack.infer(x_test_for_attack, pred=attack_x_test_predictions, values=values)\n",
"# check accuracy\n",
"train_acc = np.sum(inferred_train_bb == np.around(attack_x_test_feature, decimals=8).reshape(1,-1)) / len(inferred_train_bb)\n",
"print(train_acc)"
Expand All @@ -151,20 +152,20 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"This means that for 56% of the training set, the attacked feature is inferred correctly using this attack.\n",
"This means that for 74% of the training set, the attacked feature is inferred correctly using this attack.\n",
"Now let's check the precision and recall:"
]
},
{
"cell_type": "code",
"execution_count": 16,
"execution_count": 24,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"(0.5483870967741935, 0.32075471698113206)\n"
"(0.5816326530612245, 0.9661016949152542)\n"
]
}
],
Expand Down Expand Up @@ -205,14 +206,14 @@
},
{
"cell_type": "code",
"execution_count": 17,
"execution_count": 25,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"0.5585585585585585\n"
"0.6666666666666666\n"
]
}
],
Expand All @@ -224,7 +225,7 @@
"# train attack model\n",
"baseline_attack.fit(attack_x_train)\n",
"# infer values\n",
"inferred_train_baseline = baseline_attack.infer(attack_x_test, values=values)\n",
"inferred_train_baseline = baseline_attack.infer(x_test_for_attack, values=values)\n",
"# check accuracy\n",
"baseline_train_acc = np.sum(inferred_train_baseline == np.around(attack_x_test_feature, decimals=8).reshape(1,-1)) / len(inferred_train_baseline)\n",
"print(baseline_train_acc)"
Expand All @@ -234,13 +235,85 @@
"cell_type": "markdown",
"metadata": {},
"source": [
"In this case, the black-box attack does not do better than the baseline."
"In this case, the black-box attack does significantly better than the baseline."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## Attacking a numerical feature\n",
"Now we will try to infer the bmi level feature."
]
},
{
"cell_type": "code",
"execution_count": 29,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"54.80737471036833\n"
]
}
],
"source": [
"attack_feature = 3 # bmi\n",
"\n",
"# only attacked feature\n",
"attack_x_test_feature = attack_x_test[:, attack_feature].copy().reshape(-1, 1)\n",
"# training data without attacked feature\n",
"x_test_for_attack = np.delete(attack_x_test, attack_feature, 1)\n",
"\n",
"bb_attack = AttributeInferenceBlackBox(art_regressor, attack_feature=attack_feature)\n",
"\n",
"# train attack model\n",
"bb_attack.fit(attack_x_train)\n",
"\n",
"inferred_train_bb = bb_attack.infer(x_test_for_attack, pred=attack_x_test_predictions)\n",
"# check MSE\n",
"train_acc = np.sum((attack_x_test_feature - inferred_train_bb) ** 2) / len(inferred_train_bb)\n",
"print(train_acc)"
]
},
{
"cell_type": "code",
"execution_count": 30,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"67.66769489356126\n"
]
}
],
"source": [
"baseline_attack = AttributeInferenceBaseline(attack_feature=attack_feature)\n",
"\n",
"# train attack model\n",
"baseline_attack.fit(attack_x_train)\n",
"# infer values\n",
"inferred_train_baseline = baseline_attack.infer(x_test_for_attack)\n",
"# check MSE\n",
"baseline_train_acc = np.sum((attack_x_test_feature - inferred_train_baseline) ** 2) / len(inferred_train_baseline)\n",
"print(baseline_train_acc)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"The attack succeeds better than the baseline (a lower MSE means higher accuracy)."
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
Expand All @@ -254,7 +327,7 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.8.3"
"version": "3.9.6"
}
},
"nbformat": 4,
Expand Down