A part of ongoing university project called Research methodology and scientific writing. This project aims at exploring different spectrums of differential privacy and come up with an approach to make a good solution to handle differential privacy. The basis of our project is Google's research 'Deep Learning with Differential Privacy'. You can find the paper here, 'https://arxiv.org/abs/1607.00133'.
To implement deep learning models to perform differential privacy to withhold personal information, but also make sure the resulting data comes from the same distribution.
Differential privacy can be attained by using deep learning models, by clipping the gradient in the layers of the network.
The purpose is to use a deep learning model to attain differential privacy, thereby encrypting the personal data from reaching out to the hands of unauthorised companies and organisations.
https://github.com/google/differential-privacy/
https://github.com/Google/private-join-and-compute
Code base forked from : https://github.com/tensorflow/privacy using mnist_scratch