Relative Preferences

Structured prediction from bandit feed- back describes a learning scenario where instead of having access to a gold standard structure, a learner only receives partial feedback in form of the …...