Learning Maximum-A-Posteriori Perturbation Models for Structured Prediction in Polynomial Time

Open in new window