A preference model for structured supervised learning tasks

The preference model introduced in this paper gives a natural framework and a principled solution for a broad class of supervised learning problems with structured pre-dictions, such as predicting orders (label and instance rank-ing), and predicting rates (classification and ordinal regres-sion). We show how all these problems can be cast as linear problems in an augmented space, and we propose an on-line method to efficiently solve them. Experiments on an ordinal regression task confirm the effectiveness of the approach.