Universal Learning over Related Distributions and Adaptive Graph Transduction

The basis assumption “training and test data drawn from the same distribution” is often violated. We propose one common solution to cover various scenarios of learning under “different but related dis