TensorFlow Privacy: Learning with differential privacy for training data
Who is this presentation for?
- ML practitioners, ML researchers, and decision makers
Level
Description
When evaluating ML models, it can be difficult to tell the difference between what the models learned to generalize from training and what the models have simply memorized. And that difference can be crucial in some ML tasks, such as when ML models are trained using sensitive data. Recently, new techniques have emerged for differentially private training of ML models, including deep neural networks (DNNs), that used modified stochastic gradient descent to provide strong privacy guarantees for training data.
Those techniques are now available, and they’re both practical and can be easy to use. This said, they come with their own set of hyperparameters that need to be tuned, and they necessarily make learning less sensitive to outlier data in ways that are likely to slightly reduce utility. Úlfar Erlingsson explores the basics of ML privacy, introduces differential privacy and why it’s considered a gold standard, explains the concrete use of ML privacy and the principled techniques behind it, and dives into intended and unintended memorization and how it differs from generalization.
Prerequisite knowledge
- Experience using TensorFlow to train ML models
- A basic understanding of stochastic gradient descent
What you'll learn
- Learn what it means to provide privacy guarantees for ML models and how such guarantees can be achieved in practice using TensorFlow Privacy
Ulfar Erlingsson
Google Brain
Úlfar Erlingsson is a research scientist on the Brain team at Google, working primarily on privacy and security of deep learning systems. Previously, Úlfar led computer security research at Google and was a researcher at Microsoft Research, associate professor at Reykjavik University, cofounder and CTO of the internet security startup GreenBorder Technologies, and director of privacy protection at deCODE genetics. Úlfar holds a PhD in computer science from Cornell University.
Presented by
Diamond Sponsor
Elite Sponsors
Gold Sponsor
Supporting Sponsors
Premier Exhibitors
Exhibitors
Innovators
Contact us
confreg@oreilly.com
For conference registration information and customer service
partners@oreilly.com
For more information on community discounts and trade opportunities with O’Reilly conferences
sponsorships@oreilly.com
For information on exhibiting or sponsoring a conference
pr@oreilly.com
For media/analyst press inquires