top-k-identification

This paper addresses the fixed-confidence top-k identification problem from noisy pairwise comparisons, and develops an asymptotically optimal algorithm that minimizes the expected number of comparisons.