How does promoting the minority fraction affect generalization? A theoretical study of the one-hidden-layer neural network on group imbalance

arXiv:2403.07310v1 Announce Type: cross
Abstract: Group imbalance has been a known problem in empirical risk minimization (ERM), where the achieved high average accuracy is accompanied by low accuracy in a minority group. Despite algorithmic efforts to improve the minority group accuracy, a theoretical generalization analysis of ERM on individual groups remains elusive. By formulating the group imbalance problem with the Gaussian Mixture Model, this paper quantifies the impact of individual groups on the sample complexity, the convergence rate, and the …

abstract accuracy arxiv cs.lg erm hidden layer low network neural network risk stat.ml study type