Unraveling the Key Components of OOD Generalization via Diversification

arXiv:2312.16313v3 Announce Type: replace
Abstract: Supervised learning datasets may contain multiple cues that explain the training set equally well, i.e., learning any of them would lead to the correct predictions on the training data. However, many of them can be spurious, i.e., lose their predictive power under a distribution shift and consequently fail to generalize to out-of-distribution (OOD) data. Recently developed "diversification" methods (Lee et al., 2023; Pagliardini et al., 2023) approach this problem by finding multiple diverse hypotheses that …

abstract arxiv components cs.lg data datasets distribution diversification however key multiple power predictions predictive set shift supervised learning the key them training training data type via