What importance weights correct covariate shift?

Question: Assume covariate shift: the source and target domains have different input distributions, but Pₛ(y|x)=Pₜ(y|x). Which sample weight makes source risk resemble target risk? A) Pₛ(x)/Pₜ(x) B) Pₜ(x)/Pₛ(x) C) Pₜ(y)/Pₛ(y) D) Pₜ(y|x)/Pₛ(y|x) Correct: B Explanation: Source observations should be weighted according to how prevalent their feature values are in the target distribution relative to the source distribution. The ideal density ratio is therefore Pₜ(x)/Pₛ(x). Topic: advanced ML / domain adaptation / covariate shift