Module: tf_privacy.v1

TensorFlow Privacy library v1 imports.

This module includes classes designed to be compatible with TF1, based on tf.compat.v1.train.Optimizer and tf.estimator.Estimator.

Classes

class DNNClassifierV1: DP version of tf.compat.v1.estimator.DNNClassifier.

class DPAdagradGaussianOptimizer: DP subclass of tf.compat.v1.train.AdagradOptimizer.

class DPAdagradOptimizer: Differentially private subclass of tf.compat.v1.train.AdagradOptimizer.

class DPAdamGaussianOptimizer: DP subclass of tf.compat.v1.train.AdamOptimizer.

class DPAdamOptimizer: Differentially private subclass of tf.compat.v1.train.AdamOptimizer.

class DPGradientDescentGaussianOptimizer: DP subclass of tf.compat.v1.train.GradientDescentOptimizer.

class DPGradientDescentOptimizer: Differentially private subclass of tf.compat.v1.train.GradientDescentOptimizer.

class VectorizedDPAdagrad: Vectorized DP subclass of tf.compat.v1.train.AdagradOptimizer using Gaussian averaging.

class VectorizedDPAdagradOptimizer: Vectorized DP subclass of tf.compat.v1.train.AdagradOptimizer using Gaussian averaging.

class VectorizedDPAdam: Vectorized DP subclass of tf.compat.v1.train.AdamOptimizer using Gaussian averaging.

class VectorizedDPAdamOptimizer: Vectorized DP subclass of tf.compat.v1.train.AdamOptimizer using Gaussian averaging.

class VectorizedDPSGD: Vectorized DP subclass of tf.compat.v1.train.GradientDescentOptimizer using Gaussian averaging.

class VectorizedDPSGDOptimizer: Vectorized DP subclass of tf.compat.v1.train.GradientDescentOptimizer using Gaussian averaging.

Functions

make_optimizer_class(...): Given a subclass of tf.compat.v1.train.Optimizer, returns a DP-SGD subclass of it.

make_vectorized_optimizer_class(...): Given a subclass of tf.compat.v1.train.Optimizer, returns a vectorized DP-SGD subclass of it.