Spring 2019

Demo of Differentially Private Machine Learning via Tensorflow (2nd Floor Lounge)

Wednesday, January 30th, 2019 1:00 pm2:00 pm

Add to Calendar


Steve Chien (Google)

A brief tutorial and demo on using TF Privacy, Tensorflow's open-source library for differentially private stochastic gradient descent (DP-SGD). We will show how to easily write code that implements DP-SGD, talk about practical considerations in training, and show how to analyze the privacy parameters of the resulting model.