Smartphone Privacy: How to Learn from Distributed, Private Data

Join Jelani Nelson from University of California, Berkeley as he explores the paradox of improving smartphone features while protecting user privacy in this 50-minute talk from the 2024 SIAM Conference on Mathematics of Data Science. Discover how machine-learning algorithms that power features like auto-complete and spell check can learn from aggregate user data while providing mathematical guarantees for individual privacy—a surprising solution discovered about 20 years ago. Examine the latest algorithms that enable learning from distributed, private data without requiring users to share personal text messages with device manufacturers. This presentation offers valuable insights at the intersection of smartphone security, machine learning, and mathematical data science.