1 paper across 1 session
We introduce a model of prediction with limited selectivity, and prove instance-dependent bounds on the optimal error rate.
