The Effects of Mixed Sample Data Augmentation are Class Dependent

arXiv:2307.09136v2 Announce Type: replace
Abstract: Mixed Sample Data Augmentation (MSDA) techniques, such as Mixup, CutMix, and PuzzleMix, have been widely acknowledged for enhancing performance in a variety of tasks. A previous study reported the class dependency of traditional data augmentation (DA), where certain classes benefit disproportionately compared to others. This paper reveals a class dependent effect of MSDA, where some classes experience improved performance while others experience degraded performance. This research addresses the issue of class dependency in MSDA and …

abstract arxiv augmentation benefit class cs.ai cs.cv data effects mixed performance sample study tasks type