[Abstract]: We consider a distributed framework where training and test samples drawn from the same distribution are available, with the training instances spread across disjoint nodes. In this setting, a novel learning ...