[Invited Talk] Privacy in Metalearning and Multitask Learning: Modeling and Separations
Dates
2025-02-12 14:30 - 16:00
Venue
Artemidos 1 - Amphitheater
Title:
Privacy in Metalearning and Multitask Learning: Modeling and Separations
Speaker: Konstantina Bairaktari, PhD student in Khoury College of Computer Sciences at Northeastern University
Abstract: Model personalization allows a
set of individuals, each facing a different learning task, to train models that are more accurate for each person than those they could develop individually. The goals of personalization are captured in a variety of formal frameworks, such as multitask learning
and metalearning. Combining data for model personalization poses risks for
privacy because the output of an individual's model can depend on the data of other individuals. In this work we undertake a
systematic study of differentially private personalized learning. Our first main contribution is to construct a taxonomy of formal frameworks for private personalized learning. This taxonomy captures different formal frameworks for learning as well as different
threat models for the attacker. Our second main contribution is to prove separations between the personalized learning problems corresponding to different choices. In particular, we prove a novel separation between private multitask learning and private metalearning.
This talk is based on joint work with Maryam Aliakbarpour, Adam Smith, Marika Swanberg and Jonathan Ullman.
SHORT BIO: Konstantina is
a PhD student in Khoury College of Computer Sciences at Northeastern University, where she is advised by Huy Le Nguyen and Jonathan Ullman. Before joining Northeastern, she received her diploma in Electrical and Computer Engineering from the National Technical
University of Athens. Her research interests are in theoretical approaches to fairness in machine learning and data
privacy.
You can also attend online through MS-Teams by following the link below
