Infer membership status of samples from the target estimator. This method should be overridden by all concrete inference attack imple

Inspired by https://stackoverflow.com/questions/66695251/define-manually-sorted-mnist-dataset-with-batch-size-1-in-pytorch

ReportNoisyMax mechanism with noise_eps with 2*noise_eps-DP in our setting where one count can go up by one and another can go down by 1.

ReportNoisyMax mechanism with noise_eps with 2*noise_eps-DP in our setting where one count can go up by one and another can go down by 1.

Converts list of torch tensors to list of integers/floats. Fix for not having the functionality which converts list of tensors to tensors

Convert an array of labels to binary class matrix. :param labels: An array of integer labels of shape `(nb_samples,)`. :param nb_classes

return starting index for different labels in the sorted dataset

Computes the logmgf value given q and privacy eps. The bound used is the min of three terms. The first term is from https://arxiv.org/pd

Computes the logmgf value given q and privacy eps. The bound used is the min of three terms. The first term is from https://arxiv.org/pd

Return sensitivity at distance k. Args: counts: an array of scores noise_eps: noise parameter used l: moment whose s

Return sensitivity at distane k. Args: counts: tensor of scores noise_eps: noise parameter used l: moment whose sens

train the aia classifier to infer sensitive attribute from embeddings

Calculate metrics for each class of the train (shadow) or test (target) dataset

Returns ~ Pr[outcome != winner]. Args: counts: a list of scores noise_eps: privacy parameter for noisy_max Returns:

Returns ~ Pr[outcome != winner]. Args: counts: a list of scores noise_eps: privacy parameter for noisy_max Returns:

Function to create data loaders for the Teacher classifier

a little bit redundant since we make the data into torch dataset, but reverse them back into the original data...

Compute the mixup data. Return mixed inputs, pairs of targets, and lambda

Performs PATE analysis on predictions from teachers and combined predictions for student. Args: teacher_preds: a numpy array of dim