Convex Calibrated Surrogates for Low-Rank Loss Matrices with Applications to Subset Ranking Losses

Open in new window