2 code implementations • NeurIPS 2019 • Sunil Thulasidasan, Gopinath Chennupati, Jeff Bilmes, Tanmoy Bhattacharya, Sarah Michalak
In this work, we discuss a hitherto untouched aspect of mixup training -- the calibration and predictive uncertainty of models trained with mixup.
Ranked #1 on Out-of-Distribution Detection on STL-10