package testing

Get desktop application:
View/edit binary Protocol Buffers messages

Used in: GeneralSamplingParameters

• UNDEFINED_AGGREGATION = 0

• COUNT = 1

• SUM = 2

• MEAN = 3

• STDV = 5

• VARIANCE = 6

• QUANTILE = 7

• SELECT_PARTITIONS = 9

Used in: ApproximateBoundsDpTestCaseCollection

• optional string name = 1

• optional DpTestParameters dp_test_parameters = 2

• optional ApproximateBoundsSamplingParameters approximate_bounds_sampling_parameters = 3

• repeated ApproximateBoundsDpTestCase approximate_bounds_dp_test_case = 1

• optional VotingParameters voting_parameters = 2

• optional DpTestValidityParameters validity_parameters = 3

Parameters required for drawing sample bounds from the approximate bounds calculator of the DP library.

Used in: ApproximateBoundsDpTestCase

• optional int32 number_of_samples = 1

• optional NoiseType noise_type = 2

• optional int32 max_contributions = 3

• optional int32 number_of_bins = 4

• optional double scale = 5

• optional double base = 6

• optional double epsilon = 7

• optional double delta = 8

• repeated double raw_entry = 9

• repeated double neighbour_raw_entry = 10

Used in: BoundedMeanDpTestCaseCollection

• optional string name = 1

• optional DpTestParameters dp_test_parameters = 2

• optional BoundedMeanSamplingParameters bounded_mean_sampling_parameters = 3

• repeated BoundedMeanDpTestCase bounded_mean_dp_test_case = 1

• optional VotingParameters voting_parameters = 2

• optional DpTestValidityParameters validity_parameters = 3

Parameters required for drawing samples from the bounded mean aggregation of the DP library.

Used in: BoundedMeanDpTestCase

• optional int32 number_of_samples = 1

• optional NoiseType noise_type = 2

• optional int32 max_contributions_per_partition = 3

• optional int32 max_partitions_contributed = 4

• optional double lower_bound = 5

• optional double upper_bound = 6

• optional double epsilon = 7

• optional double delta = 8

• repeated double raw_entry = 9

• repeated double neighbour_raw_entry = 10

Used in: BoundedQuantilesDpTestCaseCollection

• optional string name = 1

• optional DpTestParameters dp_test_parameters = 2

• optional BoundedQuantilesSamplingParameters bounded_quantiles_sampling_parameters = 3

• repeated BoundedQuantilesDpTestCase bounded_quantiles_dp_test_case = 1

• optional VotingParameters voting_parameters = 2

• optional DpTestValidityParameters validity_parameters = 3

Parameters required for drawing samples from the bounded quantiles aggregation of the DP library.

Used in: BoundedQuantilesDpTestCase

• optional int32 number_of_samples = 1

• optional NoiseType noise_type = 2

• repeated double rank = 3

• optional int32 max_contributions_per_partition = 4

• optional int32 max_partitions_contributed = 5

• optional double lower_bound = 6

• optional double upper_bound = 7

• optional int32 tree_height = 8

• optional int32 branching_factor = 9

• optional double epsilon = 10

• optional double delta = 11

• repeated double raw_entry = 12

• repeated double neighbour_raw_entry = 13

Used in: BoundedStdvDpTestCaseCollection

• optional string name = 1

• optional DpTestParameters dp_test_parameters = 2

• optional BoundedStdvSamplingParameters bounded_stdv_sampling_parameters = 3

• repeated BoundedStdvDpTestCase bounded_stdv_dp_test_case = 1

• optional VotingParameters voting_parameters = 2

• optional DpTestValidityParameters validity_parameters = 3

Parameters required for drawing samples from the bounded stdv and bounded variance aggregation of the DP library.

Used in: BoundedStdvDpTestCase

• optional int32 number_of_samples = 1

• optional NoiseType noise_type = 2

• optional int32 max_contributions_per_partition = 3

• optional int32 max_partitions_contributed = 4

• optional double lower_bound = 5

• optional double upper_bound = 6

• optional double epsilon = 7

• optional double delta = 8

• repeated double raw_entry = 9

• repeated double neighbour_raw_entry = 10

Used in: BoundedSumDpTestCaseCollection

• optional string name = 1

• optional DpTestParameters dp_test_parameters = 2

• optional BoundedSumSamplingParameters bounded_sum_sampling_parameters = 3

• repeated BoundedSumDpTestCase bounded_sum_dp_test_case = 1

• optional VotingParameters voting_parameters = 2

• optional DpTestValidityParameters validity_parameters = 3

Parameters required for drawing samples from the bounded sum aggregation of the DP library.

Used in: BoundedSumDpTestCase

• optional int32 number_of_samples = 1

• optional NoiseType noise_type = 2

• optional int32 max_partitions_contributed = 3

• optional double lower_bound = 4

• optional double upper_bound = 5

• optional double epsilon = 6

• optional double delta = 7

• repeated double raw_entry = 8

• repeated double neighbour_raw_entry = 9

Parameters required for running a closeness test

Used in: DistributionClosenessTestCase

• optional double mean = 1

  mean of the distribution that is used as a reference

• optional double variance = 2

  variance of the distribution that is used as a reference

• optional double l2_tolerance = 3

  l2 distance deemed acceptable between the distribution that is being tested for and the reference distribution

• optional double granularity = 4

  parameter quantifying the amount of discretization that is applied to samples before testing

Parameters quantifying notions of specificity that a particular closeness test should satisfy.

Used in: DistributionClosenessTestCaseCollection

• optional double shift_specificity = 1

  The minimum difference between the mean of two sets of samples relative to their standard deviation that should be rejected.

• optional double scale_specificity = 2

  The minimum difference in scale between the variance of two sets of samples that should be rejected.

Used in: CountDpTestCaseCollection

• optional string name = 1

• optional DpTestParameters dp_test_parameters = 2

• optional CountSamplingParameters count_sampling_parameters = 3

• repeated CountDpTestCase count_dp_test_case = 1

• optional VotingParameters voting_parameters = 2

• optional DpTestValidityParameters validity_parameters = 3

Parameters required for drawing samples from the count aggregation of the DP library.

Used in: CountDpTestCase

• optional int32 number_of_samples = 1

• optional NoiseType noise_type = 2

• optional int32 max_partitions_contributed = 3

• optional double epsilon = 4

• optional double delta = 5

• repeated int64 raw_increment_by = 6

• repeated int64 neighbour_raw_increment_by = 7

Used in: GeneralSamplingParameters

• optional int32 user = 1

• optional int32 partition = 2

• optional double double_value = 3

Used in: DistributionClosenessTestCaseCollection

• optional string name = 1

• optional ClosenessTestParameters closeness_test_parameters = 2

• optional NoiseSamplingParameters noise_sampling_parameters = 3

• repeated DistributionClosenessTestCase distribution_closeness_test_case = 1

• optional VotingParameters voting_parameters = 2

• optional ClosenessTestValidityParameters validity_parameters = 3

Used in: DistributionDpTestCaseCollection

• optional string name = 1

• optional DpTestParameters dp_test_parameters = 2

• optional NoiseSamplingParameters noise_sampling_parameters = 3

• repeated DistributionDpTestCase distribution_dp_test_case = 1

• optional VotingParameters voting_parameters = 2

• optional DpTestValidityParameters validity_parameters = 3

Parameters required for running an approximate DP test

Used in: ApproximateBoundsDpTestCase, BoundedMeanDpTestCase, BoundedQuantilesDpTestCase, BoundedStdvDpTestCase, BoundedSumDpTestCase, CountDpTestCase, DistributionDpTestCase, LongBoundedSumDpTestCase

• optional double epsilon = 1

  parameter defining the epsilon the statistical DP test tries to verify (this epsilon should not be used for drawing samples; the epsilon for drawing samples is defined in the respective sampling proto messages. However, since we normally try to verify the value of epsilon that was used while sampling, it is common for these two epsilons to be set to the same value.)

• optional double delta = 2

  parameter defining the delta the statistical DP test tries to verify (this delta should not be used for drawing samples; the delta for drawing samples is defined in the respective sampling proto messages)

• optional double delta_tolerance = 3

  parameter specifying the acceptable deviation from delta

• optional double granularity = 4

  parameter quantifying the amount of discretization that is applied to samples before testing

• optional int32 num_of_buckets = 5

  parameter defining the number of buckets to which samples are mapped before testing

Parameters quantifying notions of specificity that a particular approximate DP test should satisfy. These parameters control the behaviour of the validity tests: they specify which inputs we expect the statistical tests to reject. Setting the parameters too low will cause the validity tests to fail (because our statistical tests aren't sensitive enough to notice the difference) and setting them too high will cause the validity tests to pass even if our statistical tests can't spot when the algorithm is not DP.

Used in: ApproximateBoundsDpTestCaseCollection, BoundedMeanDpTestCaseCollection, BoundedQuantilesDpTestCaseCollection, BoundedStdvDpTestCaseCollection, BoundedSumDpTestCaseCollection, CountDpTestCaseCollection, DistributionDpTestCaseCollection, LongBoundedSumDpTestCaseCollection

• optional double distance_specificity = 1

  The minimum distance between two sets of samples relative to their anticipated distance that should be rejected.

• optional double failure_specificity = 2

  The minimum rate of privacy failures relative to the delta tolerance of the DP test that should be rejected.

• optional double epsilon_specificity = 3

  The minimum ratio between an epsilon that should be rejected and an epsilon that should be accepted, i.e., if epsilon is the smallest test parameter for which we expect the test to accept, then a test parameter of epsilon * epsilon_specificity should be rejected.

• optional double delta_specificity = 4

  The minimum ratio between a delta that should be rejected and a delta that should be accepted, i.e., if delta is the test privacy parameter for which which we expect the test to accept, a test parameter of delta * delta_specificity should be rejected.

General parameters for drawing samples, intended for testing various functionality of higher level tools.

• optional int32 number_of_samples = 1

• repeated AggregationType aggregation_type = 2

• optional int32 max_partitions_contributed = 3

• optional int32 max_contributions_per_partition = 4

• optional double lower_bound = 5

• optional double upper_bound = 6

• optional double epsilon = 7

• optional double delta = 8

• repeated DataRecord record = 9

• repeated DataRecord neighbour_record = 10

Used in: LongBoundedSumDpTestCaseCollection

• optional string name = 1

• optional DpTestParameters dp_test_parameters = 2

• optional LongBoundedSumSamplingParameters long_bounded_sum_sampling_parameters = 3

• repeated LongBoundedSumDpTestCase long_bounded_sum_dp_test_case = 1

• optional VotingParameters voting_parameters = 2

• optional DpTestValidityParameters validity_parameters = 3

Parameters required for drawing samples from the long bounded sum aggregation of the DP library.

Used in: LongBoundedSumDpTestCase

• optional int32 number_of_samples = 1

• optional NoiseType noise_type = 2

• optional int32 max_partitions_contributed = 3

• optional int64 lower_bound = 4

• optional int64 upper_bound = 5

• optional double epsilon = 6

• optional double delta = 7

• repeated int64 raw_entry = 8

• repeated int64 neighbour_raw_entry = 9

Parameters required for drawing samples from the add noise method of the DP library.

Used in: DistributionClosenessTestCase, DistributionDpTestCase

• optional int32 number_of_samples = 1

• optional int32 l0_sensitivity = 2

• optional double linf_sensitivity = 3

• optional double epsilon = 4

• optional double delta = 5

• optional double raw_input = 6

Used in: ApproximateBoundsSamplingParameters, BoundedMeanSamplingParameters, BoundedQuantilesSamplingParameters, BoundedStdvSamplingParameters, BoundedSumSamplingParameters, CountSamplingParameters, LongBoundedSumSamplingParameters

• UNDEFINED_NOISE = 0

• LAPLACE = 1

• GAUSSIAN = 2

Parameters specifying a voting mechanism for evaluating multiple runs of a particular statistical test. This is intended to reduce the probability of error compared to evaluating just a single run of the respective test.

Used in: ApproximateBoundsDpTestCaseCollection, BoundedMeanDpTestCaseCollection, BoundedQuantilesDpTestCaseCollection, BoundedStdvDpTestCaseCollection, BoundedSumDpTestCaseCollection, CountDpTestCaseCollection, DistributionClosenessTestCaseCollection, DistributionDpTestCaseCollection, LongBoundedSumDpTestCaseCollection

• optional int32 number_of_votes = 1