Where Does My Model Underperform? A Human Evaluation of Slice Discovery Algorithms

Machine learning (ML) models that achieve high average accuracy can still underperform on semantically coherent subsets ("slices") of data. This behavior can have significant societal consequences for the safety or bias of the model in deployment, but identifying these underperforming slices can be difficult in practice, especially in domains where practitioners lack access to group annotations to define coherent subsets of their data. Motivated by these challenges, ML researchers have developed new slice discovery algorithms that aim to group together …

accuracy algorithms behavior bias consequences cs.cv cs.hc cs.lg data deployment discovery evaluation human machine machine learning practice safety