MEMR: A Margin Equipped Monotone Retargeting Framework for Ranking

We bring to bear the tools of convexity, margins and the newly proposed technique of monotone retargeting upon the task of learning permutations from examples. This leads to novel and efficient algorithms with guaranteed prediction performance in the online setting and on global optimality and the rate of convergence in the batch setting. Monotone retargeting efficiently optimizes over all possible monotone transformations as well as the finite dimensional parameters of the model. As a result we obtain an effective algorithm to learn transitive relationships over items. It captures the inherent combinatorial characteristics of the output space yet it has a computational burden not much more than that of a generalized linear model.