Identifiability of Priors from Bounded Sample Sizes with Applications to Transfer Learning

We explore a transfer learning setting, in which a finite sequence of target concepts are sampled independently with an unknown distribution from a known family. We study the total number of labeled