In this talk, I will aim to cover two problems in fairness-aware learning. One is an offline classification problem, where data are distributed and privacy constraints are required. We will show how fairness in terms of statistical parity changes the statistical learning accuracy, and how it changes the cost of privacy. This part of the talk is based on the arXiv preprint 2603.24392. The other is based on some ongoing work, where we study how imposing individual fairness constraints essentially turns a parametric problem (contextual multi-armed bandits) to a nonparametric problem.
Yi Yu 于怡 : Two problems in fairness-aware learning
Yi Yu 于怡 is Professor in Statistics and the Deputy Head of Department (Research) in the Department of Statistics, University of Warwick. She is interested in various aspects of high-dimensional and nonparametric statistics, with a recent interest in trustworthy AI related problems. She has worked in high-dimensional regression, differential privacy, change point analysis, statistical network analysis, nonparametric statistics and functional data analysis.
Published Sep. 16, 2026 2:13 PM
- Last modified Sep. 16, 2026 2:13 PM