Explore federated heavy hitters discovery with differential privacy in this 44-minute lecture by Peter Kairouz from Google AI. Delve into the intersection of privacy and data analysis, covering key topics such as federated learning, multiparty computation, and open research. Examine two models and four essential ingredients of the algorithm, along with its parameters and results. Analyze tradeoffs, deltas, and comparisons to gain a comprehensive understanding of this cutting-edge approach to privacy-preserving data analysis in federated settings.
Federated Heavy Hitters Discovery with Differential Privacy
Overview
Syllabus
Introduction
Federated Learning
Multiparty Computation
Open Research
Applications
Differential Privacy
Two Models
Four Essential Ingredients
The Algorithm
Algorithm Parameters
Results
Tradeoffs
Deltas
Comparison
Taught by
Simons Institute
Related Courses
-
Secure and Private AI
-
Build and operate machine learning solutions with Azure Machine Learning
-
Privacy-Aware Compression for Federated Learning
-
Preserving Data Privacy in Federated Learning - Xiaokui Xiao
4.0 -
Federated Learning with Formal User-Level Differential Privacy Guarantees
-
Federated Learning and Analytics at Google - Workshop Introduction
