Simple and near-optimal algorithms for hidden stratification and multi-group learning
arXiv:2112.12181v229 citations
Originality Incremental advance
AI Analysis
This work addresses fairness and stratification issues in machine learning, offering practical algorithms for multi-group scenarios.
The paper tackled the multi-group learning problem, which addresses subgroup fairness and hidden stratification, by studying solution structures and providing simple, near-optimal algorithms.
Multi-group agnostic learning is a formal learning criterion that is concerned with the conditional risks of predictors within subgroups of a population. The criterion addresses recent practical concerns such as subgroup fairness and hidden stratification. This paper studies the structure of solutions to the multi-group learning problem, and provides simple and near-optimal algorithms for the learning problem.