Fairness Through Robustness - Investigating Robustness Disparity in Deep Learning

Explore a 21-minute conference talk that delves into the intersection of fairness and robustness in deep learning systems. Investigate how robustness disparity affects different demographic groups and its implications for algorithmic fairness. Learn about data preliminaries, metrics, and key research questions addressed in this study. Examine the results obtained from experiments on various datasets, including the intriguing "Aliens" data. Discover potential mitigation strategies to address robustness disparities and enhance fairness in machine learning models. Gain insights into the critical relationship between robustness and fairness in AI systems, presented by researchers V. Nanda, S. Dooley, S. Singla, S. Feizi, and J. Dickerson at the FAccT 2021 virtual conference.

Introduction
Data Preliminaries
Metrics
Research Questions
Results
Aliens Data
Mitigation