Examining and Combating Spurious Features under Distribution Shift
Chunting Zhou 1 Xuezhe Ma 2 Paul Michel 1 Graham Neubig 1
(a) ERM (b) Group DRO [clean]
(average=93.3% robust=83.4%) (average=96.5% robust=95.8%)
Abstract
A central goal of machine learning is to learn
robust representations that capture the causal rela-
tionship between inputs features and output l ...


雷达卡




京公网安备 11010802022788号







