Adding comparison with other uncoupled regression: Since Reviewers 1 and 4 share the concern on the SVMRank

We thank the reviewers for their thoughtful and useful feedback. We will also fix minor typos in the final version of the paper. However, we may obtain some linear model by using the methods discussed in Hsu et al. As you suspect, the discussion should have gone for Figure 2. We will fix it in the final version. It is because we used log-plot in figures.