def test_complex_nested_query(self):
with self.cached_session() as sess:
query_ab = gaussian_query.GaussianSumQuery(l2_norm_clip=1.0,
stddev=0.0)
query_c = gaussian_query.GaussianAverageQuery(l2_norm_clip=10.0,
sum_stddev=0.0,
denominator=2.0)
query_d = gaussian_query.GaussianSumQuery(l2_norm_clip=10.0,
stddev=0.0)
query = nested_query.NestedQuery(
[query_ab, {
'c': query_c,
'd': [query_d]
}])
record1 = [{
'a': 0.0,
'b': 2.71828
}, {
'c': (-4.0, 6.0),
'd': [-4.0]
}]
record2 = [{
'a': 3.14159,
'b': 0.0
}, {
'c': (6.0, -4.0),
'd': [5.0]
}]
query_result = _run_query(query, [record1, record2])
result = sess.run(query_result)
expected = [{'a': 1.0, 'b': 1.0}, {'c': (1.0, 1.0), 'd': [1.0]}]
self.assertAllClose(result, expected)
def testBaseline(self, cls, num_microbatches, expected_answer):
with self.cached_session() as sess:
var0 = tf.Variable([1.0, 2.0])
data0 = tf.Variable([[3.0, 4.0], [5.0, 6.0], [7.0, 8.0],
[-1.0, 0.0]])
ledger = privacy_ledger.PrivacyLedger(1e6, num_microbatches / 1e6,
50, 50)
dp_average_query = gaussian_query.GaussianAverageQuery(
1.0e9, 0.0, num_microbatches, ledger)
dp_average_query = privacy_ledger.QueryWithLedger(
dp_average_query, ledger)
opt = cls(dp_average_query,
num_microbatches=num_microbatches,
learning_rate=2.0)
self.evaluate(tf.global_variables_initializer())
# Fetch params to validate initial values
self.assertAllClose([1.0, 2.0], self.evaluate(var0))
# Expected gradient is sum of differences divided by number of
# microbatches.
gradient_op = opt.compute_gradients(loss(data0, var0), [var0])
grads_and_vars = sess.run(gradient_op)
self.assertAllCloseAccordingToType(expected_answer,
grads_and_vars[0][0])
def test_nested_query_with_noise(self):
with self.cached_session() as sess:
sum_stddev = 2.71828
denominator = 3.14159
query1 = gaussian_query.GaussianSumQuery(l2_norm_clip=1.5,
stddev=sum_stddev)
query2 = gaussian_query.GaussianAverageQuery(
l2_norm_clip=0.5,
sum_stddev=sum_stddev,
denominator=denominator)
query = nested_query.NestedQuery((query1, query2))
record1 = (3.0, [2.0, 1.5])
record2 = (0.0, [-1.0, -3.5])
query_result = _run_query(query, [record1, record2])
noised_averages = []
for _ in xrange(1000):
noised_averages.append(nest.flatten(sess.run(query_result)))
result_stddev = np.std(noised_averages, 0)
avg_stddev = sum_stddev / denominator
expected_stddev = [sum_stddev, avg_stddev, avg_stddev]
self.assertArrayNear(result_stddev, expected_stddev, 0.1)
def __init__(
self,
l2_norm_clip,
noise_multiplier,
num_microbatches,
unroll_microbatches=False,
*args, # pylint: disable=keyword-arg-before-vararg
**kwargs):
dp_average_query = gaussian_query.GaussianAverageQuery(
l2_norm_clip, l2_norm_clip * noise_multiplier, num_microbatches)
if 'population_size' in kwargs:
population_size = kwargs.pop('population_size')
max_queries = kwargs.pop('ledger_max_queries', 1e6)
max_samples = kwargs.pop('ledger_max_samples', 1e6)
selection_probability = num_microbatches / population_size
ledger = privacy_ledger.PrivacyLedger(
population_size,
selection_probability,
max_samples,
max_queries)
dp_average_query = privacy_ledger.QueryWithLedger(
dp_average_query, ledger)
super(DPGaussianOptimizerClass, self).__init__(
dp_average_query,
num_microbatches,
unroll_microbatches,
*args,
**kwargs)
def testNoiseMultiplier(self, cls):
with tf.GradientTape(persistent=True) as gradient_tape:
var0 = tf.Variable([0.0])
data0 = tf.Variable([[0.0]])
ledger = privacy_ledger.PrivacyLedger(1e6, 1 / 1e6, 5000, 5000)
dp_average_query = gaussian_query.GaussianAverageQuery(4.0, 8.0, 1)
dp_average_query = privacy_ledger.QueryWithLedger(
dp_average_query, ledger)
opt = cls(dp_average_query, num_microbatches=1, learning_rate=2.0)
self.evaluate(tf.global_variables_initializer())
# Fetch params to validate initial values
self.assertAllClose([0.0], self.evaluate(var0))
grads = []
for _ in range(1000):
grads_and_vars = opt.compute_gradients(
lambda: self._loss_fn(var0, data0), [var0],
gradient_tape=gradient_tape)
grads.append(grads_and_vars[0][0])
# Test standard deviation is close to l2_norm_clip * noise_multiplier.
self.assertNear(np.std(grads), 2.0 * 4.0, 0.5)
def test_nested_query(self):
population_size = tf.Variable(0)
selection_probability = tf.Variable(0.0)
ledger = privacy_ledger.PrivacyLedger(population_size,
selection_probability, 50, 50)
query1 = gaussian_query.GaussianAverageQuery(l2_norm_clip=4.0,
sum_stddev=2.0,
denominator=5.0,
ledger=ledger)
query2 = gaussian_query.GaussianAverageQuery(l2_norm_clip=5.0,
sum_stddev=1.0,
denominator=5.0,
ledger=ledger)
query = nested_query.NestedQuery([query1, query2])
query = privacy_ledger.QueryWithLedger(query, ledger)
record1 = [1.0, [12.0, 9.0]]
record2 = [5.0, [1.0, 2.0]]
# First sample.
tf.assign(population_size, 10)
tf.assign(selection_probability, 0.1)
test_utils.run_query(query, [record1, record2])
expected_queries = [[4.0, 2.0], [5.0, 1.0]]
formatted = ledger.get_formatted_ledger_eager()
sample_1 = formatted[0]
self.assertAllClose(sample_1.population_size, 10.0)
self.assertAllClose(sample_1.selection_probability, 0.1)
self.assertAllClose(sorted(sample_1.queries), sorted(expected_queries))
# Second sample.
tf.assign(population_size, 20)
tf.assign(selection_probability, 0.2)
test_utils.run_query(query, [record1, record2])
formatted = ledger.get_formatted_ledger_eager()
sample_1, sample_2 = formatted
self.assertAllClose(sample_1.population_size, 10.0)
self.assertAllClose(sample_1.selection_probability, 0.1)
self.assertAllClose(sorted(sample_1.queries), sorted(expected_queries))
self.assertAllClose(sample_2.population_size, 20.0)
self.assertAllClose(sample_2.selection_probability, 0.2)
self.assertAllClose(sorted(sample_2.queries), sorted(expected_queries))
def __init__(self, l2_norm_clip, noise_multiplier, num_microbatches,
*args, **kwargs):
super(DPOptimizerClass, self).__init__(*args, **kwargs)
stddev = l2_norm_clip * noise_multiplier
self._num_microbatches = num_microbatches
self._private_query = gaussian_query.GaussianAverageQuery(
l2_norm_clip, stddev, num_microbatches)
self._global_state = self._private_query.initial_global_state()
def test_nested_gaussian_average_with_clip_no_noise(self):
with self.cached_session() as sess:
query1 = gaussian_query.GaussianAverageQuery(l2_norm_clip=4.0,
sum_stddev=0.0,
denominator=5.0)
query2 = gaussian_query.GaussianAverageQuery(l2_norm_clip=5.0,
sum_stddev=0.0,
denominator=5.0)
query = nested_query.NestedQuery([query1, query2])
record1 = [1.0, [12.0, 9.0]] # Clipped to [1.0, [4.0, 3.0]]
record2 = [5.0, [1.0, 2.0]] # Clipped to [4.0, [1.0, 2.0]]
query_result = _run_query(query, [record1, record2])
result = sess.run(query_result)
expected = [1.0, [1.0, 1.0]]
self.assertAllClose(result, expected)
def test_gaussian_average_no_noise(self):
with self.cached_session() as sess:
record1 = tf.constant([5.0, 0.0]) # Clipped to [3.0, 0.0].
record2 = tf.constant([-1.0, 2.0]) # Not clipped.
query = gaussian_query.GaussianAverageQuery(
l2_norm_clip=3.0, sum_stddev=0.0, denominator=2.0)
query_result, _ = test_utils.run_query(query, [record1, record2])
result = sess.run(query_result)
expected_average = [1.0, 1.0]
self.assertAllClose(result, expected_average)
def linear_model_fn(features, labels, mode):
preds = tf.keras.layers.Dense(1, activation='linear',
name='dense').apply(features['x'])
vector_loss = tf.squared_difference(labels, preds)
scalar_loss = tf.reduce_mean(vector_loss)
dp_average_query = gaussian_query.GaussianAverageQuery(1.0, 0.0, 1)
optimizer = dp_optimizer.DPGradientDescentOptimizer(
dp_average_query, num_microbatches=1, learning_rate=1.0)
global_step = tf.train.get_global_step()
train_op = optimizer.minimize(loss=vector_loss,
global_step=global_step)
return tf.estimator.EstimatorSpec(mode=mode,
loss=scalar_loss,
train_op=train_op)
def test_gaussian_average_with_noise(self):
with self.cached_session() as sess:
record1, record2 = 2.71828, 3.14159
sum_stddev = 1.0
denominator = 2.0
query = gaussian_query.GaussianAverageQuery(
l2_norm_clip=5.0, sum_stddev=sum_stddev, denominator=denominator)
query_result, _ = test_utils.run_query(query, [record1, record2])
noised_averages = []
for _ in range(1000):
noised_averages.append(sess.run(query_result))
result_stddev = np.std(noised_averages)
avg_stddev = sum_stddev / denominator
self.assertNear(result_stddev, avg_stddev, 0.1)
def testClippingNorm(self, cls):
with self.cached_session() as sess:
var0 = tf.Variable([0.0, 0.0])
data0 = tf.Variable([[3.0, 4.0], [6.0, 8.0]])
dp_average_query = gaussian_query.GaussianAverageQuery(1.0, 0.0, 1)
opt = cls(dp_average_query, num_microbatches=1, learning_rate=2.0)
self.evaluate(tf.global_variables_initializer())
# Fetch params to validate initial values
self.assertAllClose([0.0, 0.0], self.evaluate(var0))
# Expected gradient is sum of differences.
gradient_op = opt.compute_gradients(loss(data0, var0), [var0])
grads_and_vars = sess.run(gradient_op)
self.assertAllCloseAccordingToType([-0.6, -0.8],
grads_and_vars[0][0])
def testNoiseMultiplier(self, cls):
with self.cached_session() as sess:
var0 = tf.Variable([0.0])
data0 = tf.Variable([[0.0]])
dp_average_query = gaussian_query.GaussianAverageQuery(4.0, 8.0, 1)
opt = cls(dp_average_query, num_microbatches=1, learning_rate=2.0)
self.evaluate(tf.global_variables_initializer())
# Fetch params to validate initial values
self.assertAllClose([0.0], self.evaluate(var0))
gradient_op = opt.compute_gradients(loss(data0, var0), [var0])
grads = []
for _ in range(1000):
grads_and_vars = sess.run(gradient_op)
grads.append(grads_and_vars[0][0])
# Test standard deviation is close to l2_norm_clip * noise_multiplier.
self.assertNear(np.std(grads), 2.0 * 4.0, 0.5)
def testClippingNorm(self, cls):
with tf.GradientTape(persistent=True) as gradient_tape:
var0 = tf.Variable([0.0, 0.0])
data0 = tf.Variable([[3.0, 4.0], [6.0, 8.0]])
ledger = privacy_ledger.PrivacyLedger(1e6, 1 / 1e6, 50, 50)
dp_average_query = gaussian_query.GaussianAverageQuery(1.0, 0.0, 1)
dp_average_query = privacy_ledger.QueryWithLedger(
dp_average_query, ledger)
opt = cls(dp_average_query, num_microbatches=1, learning_rate=2.0)
self.evaluate(tf.global_variables_initializer())
# Fetch params to validate initial values
self.assertAllClose([0.0, 0.0], self.evaluate(var0))
# Expected gradient is sum of differences.
grads_and_vars = opt.compute_gradients(
lambda: self._loss_fn(var0, data0), [var0],
gradient_tape=gradient_tape)
self.assertAllCloseAccordingToType([-0.6, -0.8],
grads_and_vars[0][0])
def testUnrollMicrobatches(self, cls):
with self.cached_session() as sess:
var0 = tf.Variable([1.0, 2.0])
data0 = tf.Variable([[3.0, 4.0], [5.0, 6.0], [7.0, 8.0],
[-1.0, 0.0]])
dp_average_query = (gaussian_query.GaussianAverageQuery(
1.0e9, 0.0, 4))
opt = cls(dp_average_query,
num_microbatches=4,
learning_rate=2.0,
unroll_microbatches=True)
self.evaluate(tf.global_variables_initializer())
# Fetch params to validate initial values
self.assertAllClose([1.0, 2.0], self.evaluate(var0))
# Expected gradient is sum of differences divided by number of
# microbatches.
gradient_op = opt.compute_gradients(loss(data0, var0), [var0])
grads_and_vars = sess.run(gradient_op)
self.assertAllCloseAccordingToType([-2.5, -2.5],
grads_and_vars[0][0])
