MixBag: Bag-Level Data Augmentation for Learning from Label Proportions

no code implementations • ICCV 2023 • Takanori Asanomi, Shinnosuke Matsuo, Daiki Suehiro, Ryoma Bise

In this paper, we propose a bag-level data augmentation method for LLP called MixBag, based on the key observation from our preliminary experiments; that the instance-level classification accuracy improves as the number of labeled bags increases even though the total number of instances is fixed.

Data Augmentation Weakly-supervised Learning