Dominant Arm Identification with Mixing and Recycling Observed Samples
Sample-complexity comparison in the biomedical treatment-response setting.
Proposed a novel dominance score criterion based on local distributional dominance, capturing crossing CDFs, tail behavior, and heterogeneous reward profiles that may be overlooked by mean-based comparisons.
Developed a doubly robust estimator with a joint mixing-and-recycling mechanism that guarantees simultaneous convergence of the empirical distribution functions for all arms.
Proposed the Doubly Robust Dominance Score Elimination (DRDSE) algorithm, which correctly identifies the best dominance score arm with a sample complexity upper bound that matches the lower bound up to logarithmic factors.
