no code implementations • NeurIPS 2021 • Mike Li, Hongseok Namkoong, Shangzhou Xia
The performance of ML models degrades when the training population is different from that seen under operation.
