why we use p(k|x_t, w) / λ_k in the target domain instead of only p(k|x_t, w). In my opinion, λ_k is used for select confidence threshold as in Eq.4
why we use p(k|x_t, w) / λ_k in the target domain instead of only p(k|x_t, w). In my opinion, λ_k is used for select confidence threshold as in Eq.4