Bias Amplification Enhances Minority Group Performance

arXiv:2309.06717v2 Announce Type: replace
Abstract: Neural networks produced by standard training are known to suffer from poor accuracy on rare subgroups despite achieving high accuracy on average, due to the correlations between certain spurious features and labels. Previous approaches based on worst-group loss minimization (e.g. Group-DRO) are effective in improving worse-group accuracy but require expensive group annotations for all the training samples. In this paper, we focus on the more challenging and realistic setting where group annotations are only available …

abstract accuracy arxiv bias correlations cs.cy cs.lg features improving labels loss networks neural networks performance standard subgroups training type