626_privacy/tensorflow_privacy
2
0
Fork 1
Code Issues Pull requests 1 Projects Releases Packages Wiki Activity Actions

Closes #25

Browse source 
PiperOrigin-RevId: 239031260
This commit is contained in:
Nicolas Papernot 2019-03-18 11:54:41 -07:00 committed by A. Unique TensorFlower
parent a9840529c4
commit f58891f3e3
2 changed files with 22 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

5
README.md
View file

@ -67,13 +67,14 @@ Github pull requests. To speed the code review process, we ask that:
To help you get started with the functionalities provided by this library, the
`tutorials/` folder comes with scripts demonstrating how to use the library
features.
features. The list of tutorials is described in the README included in the
tutorials directory.
NOTE: the tutorials are maintained carefully. However, they are not considered
part of the API and they can change at any time without warning. You should not
write 3rd party code that imports the tutorials and expect that the interface
will not break.
## Research directory
This folder contains code to reproduce results from research papers related to

20
tutorials/README.md
View file

@ -1,10 +1,28 @@
# Tutorials
This folder contains a set of tutorials that demonstrate the features of this
library.
As demonstrated on MNIST in `mnist_dpsgd_tutorial.py`, the easiest way to use
a differentially private optimizer is to modify an existing training loop
a differentially private optimizer is to modify an existing TF training loop
to replace an existing vanilla optimizer with its differentially private
counterpart implemented in the library.
Here is a list of all the tutorials included:
* `lm_dpsgd_tutorial.py`: learn a language model with differential privacy.
* `mnist_dpsgd_tutorial.py`: learn a convolutional neural network on MNIST with
differential privacy.
* `mnist_dpsgd_tutorial_eager.py`: learn a convolutional neural network on MNIST
with differential privacy using Eager mode.
* `mnist_dpsgd_tutorial_keras.py`: learn a convolutional neural network on MNIST
with differential privacy using tf.Keras.
The rest of this README describes the different parameters used to configure
DP-SGD as well as expected outputs for the `mnist_dpsgd_tutorial.py` tutorial.
## Parameters
All of the optimizers share some privacy-specific parameters that need to