1 paper · 1 filter
Ouyang Xu, Baoming Zhang, Ruiyu Mao +1
Deep learning models for visual recognition often exhibit systematic errors due to underrepresented semantic subpopulations. Although existing debugging frameworks can pinpoint the…