Chasing Your Long Tails - Differentially Private Prediction in Health Care Settings
Association for Computing Machinery (ACM) via YouTube
Overview
Explore the challenges and solutions of implementing differentially private machine learning in healthcare settings through this 19-minute conference talk from FAccT 2021. Delve into the unique aspects of healthcare data, examine the extreme tradeoffs involved, and understand the concept of group fairness defined by influence. Learn about differentially private training techniques, important considerations for implementation, and future directions in this field. Gain insights into how privacy-preserving methods can be applied to sensitive health data while maintaining utility and fairness in predictive models.
Syllabus
Intro
Differential Privacy
What makes ML in Health Care Different?
Datasets
Differentially Private Training
Extreme Tradeoffs in Health Care
Group Fairness Defined by Influence
Important considerations
Future Directions
Taught by
ACM FAccT Conference