The Effects of Mixed Sample Data Augmentation are Class Dependent
This addresses the issue of class dependency in MSDA for machine learning practitioners, but it is incremental as it builds on prior work on class dependency in data augmentation.
The paper tackles the problem of class-dependent performance effects in Mixed Sample Data Augmentation (MSDA), where some classes improve while others degrade, and proposes an algorithm using a mixture of MSDA and non-MSDA data to mitigate this, improving overall accuracy.
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 proposes an algorithm to mitigate it. The approach involves training on a mixture of MSDA and non-MSDA data, which not only mitigates the negative impact on the affected classes, but also improves overall accuracy. Furthermore, we provide in-depth analysis and discussion of why MSDA introduced class dependencies and which classes are most likely to have them.