An Investigation of Why Overparameterization Exacerbates Spurious Correlations
Shiori Sagawa,u00a0Aditi Raghunathan,u00a0Pang Wei Koh,u00a0Percy Liang
We study why overparameterizationu2014increasing model size well beyond the point of zero training erroru2014can hurt test error on minority groups despite improving average test error when there are spurious correlations in the data. Through simulations and experiments on two image datasets, we identify two key properties of the training data that drive this behavior: the proportions of majority versus minority groups, and the signal-to-noise ratio of the spurious correlations. We then analyze a linear setting and theoretically show how the inductive bias of models towards u201cmemorizingu201d fewer examples can cause overparameterization to hurt. Our analysis leads to a counterintuitive approach of subsampling the majority group, which empirically achieves low minority error in the overparameterized regime, even though the standard approach of upweighting the minority fails. Overall, our results suggest a tension between using overparameterized models versus using all the training data for achieving low worst-group error.
