Learning from Aggregate Responses

Explore a 45-minute lecture on learning from aggregate responses presented by Adel Javanmard from the University of Southern California at IPAM's EnCORE Workshop. Delve into the challenges and recent developments in aggregate learning frameworks, where training data is grouped into bags with aggregate responses to protect user privacy. Discover how priors can inform bag construction and learn about an iterative boosting algorithm that refines priors through sample splitting. Gain insights into loss construction and bagging schemes that enhance model accuracy while maintaining privacy. Understand the practical applications and implications of this approach in various fields where sensitive user data is involved.