Fix Privacy Bugs in Odometer and Covariance

sql/HISTORY.md

@@ -1,3 +1,7 @@
+# SmartNoise SQL v1.0.7 Release Notes
+
+* Fix odomoter accounting bug (thanks @tudorcebere!)
+
 # SmartNoise SQL v1.0.6 Release Notes
 
 * Allow newer versions of Graphviz

sql/VERSION

@@ -1 +1 @@
-1.0.6
+1.0.7

sql/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "smartnoise-sql"
-version = "1.0.6"
+version = "1.0.7"
 description = "Differentially Private SQL Queries"
 authors = ["SmartNoise Team <smartnoise@opendp.org>"]
 license = "MIT"
@@ -10,7 +10,7 @@ repository = "https://github.com/opendp/smartnoise-sdk"
 readme = "README.md"
 
 [tool.poetry.dependencies]
-python = ">=3.8,<3.14"
+python = ">=3.9,<3.14"
 opendp = ">=0.8.0,<0.13.0"
 antlr4-python3-runtime = "4.9.3"
 PyYAML = "^6.0.1"

sql/snsql/sql/odometer.py

@@ -29,7 +29,7 @@ def spent(self):
         basic = self.k * epsilon
         optimal_left_side = ((np.exp(epsilon) - 1) * epsilon * self.k)/(np.exp(epsilon) + 1)
         optimal_a = optimal_left_side + epsilon * np.sqrt(2 * self.k * np.log(epsilon + (np.sqrt(self.k*epsilon*epsilon)/tol)))
-        optimal_b = optimal_left_side + epsilon * np.sqrt(2 * self.k * (1/tol))
+        optimal_b = optimal_left_side + epsilon * np.sqrt(2 * self.k * np.log(1 / tol))
         delta = 1 - (1 - delta) ** self.k
         delta = delta * (1 - delta) + self.tol
         return tuple([min(basic, optimal_a, optimal_b), delta])

sql/tests/odometer/test_odometer.py

@@ -108,7 +108,7 @@ def test_odo_hom(self):
         for _ in range(300):
             odo.spend()
         eps, delt = odo.spent
-        assert(np.isclose(eps, 8.4917))
+        assert(np.isclose(eps, 8.2519))
         assert(np.isclose(delt, 0.19256))
     def test_odo_het(self):
         privacy = Privacy(epsilon=0.1, delta = 1/(1000))

synth/HISTORY.md

@@ -1,3 +1,7 @@
+# SmartNoise Synth v1.0.6 Release Notes
+
+* Fix privacy leak in covariance (thanks, @tudorcebere!)
+
 # SmartNoise Synth v1.0.5 Release Notes
 
 * Upgrade to OpenDP v0.12.0

synth/VERSION

@@ -1 +1 @@
-1.0.5
+1.0.6

synth/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "smartnoise-synth"
-version = "1.0.5"
+version = "1.0.6"
 description = "Differentially Private Synthetic Data"
 authors = ["SmartNoise Team <smartnoise@opendp.org>"]
 license = "MIT"
@@ -14,7 +14,7 @@ python = ">=3.9,<3.13"
 opacus = "^0.14.0"
 torch = {version = ">=2.2.0", optional = true}
 pac-synth = "^0.0.8"
-smartnoise-sql = "^1.0.5"
+smartnoise-sql = ">=1.0.7"
 Faker = ">=17.0.0"
 
 [tool.poetry.dev-dependencies]

synth/snsynth/models/dp_covariance.py

@@ -112,10 +112,10 @@ def dp_noise(n, noise_scale):
             u = np.random.uniform(size=n)
             return q_lap_iter(u, b=noise_scale)
 
-        true_val = covar(data.values.T, self.intercept)
-        scale = self.sens / self.epsilon
-        val = np.array(true_val) + dp_noise(n=len(true_val), noise_scale=scale)
-        return list(val)
+        true_val = covar(new_data.values.T, self.intercept)
+        scale = self.sens / self.epsilon
+        val = np.array(true_val) + dp_noise(n=len(true_val), noise_scale=scale)
+        return list(val)
 
     # TODO: this implementation only works for one dependent variable right now
     def get_linear_regression(self, data, x_names, y_name, intercept=False):

synth/snsynth/models/linear_regression.py

@@ -1,6 +1,9 @@
 import pandas as pd
 from sklearn.base import RegressorMixin, MultiOutputMixin
-from sklearn.linear_model.base import LinearModel
+try:
+    from sklearn.linear_model.base import LinearModel
+except ImportError:
+    from sklearn.linear_model._base import LinearModel
 from .dp_covariance import DPcovariance
 
 

synth/tests/test_dp_covariance.py

@@ -0,0 +1,22 @@
+import numpy as np
+import pandas as pd
+
+from snsynth.models.dp_covariance import DPcovariance
+
+
+def test_release_uses_clipped_data(monkeypatch):
+    # Make Laplace noise exactly zero so output equals the internal covariance.
+    monkeypatch.setattr(np.random, "uniform", lambda size: np.full(size, 0.5))
+
+    data = pd.DataFrame({"x": [2.0, 0.0], "y": [2.0, 0.0]})
+    bounds = pd.DataFrame({"x": [0.0, 1.0], "y": [0.0, 1.0]})
+
+    model = DPcovariance(n=2, cols=["x", "y"], rng=bounds, global_eps=1.0)
+    release = np.array(model.release(data))
+
+    # Expected covariance of clipped data [[1,1],[0,0]].
+    clipped = np.array([[1.0, 1.0], [0.0, 0.0]]).T
+    cov = np.cov(clipped)
+    expected = cov[np.tril_indices(cov.shape[0])]
+
+    np.testing.assert_allclose(release, expected)