Beyond Minimax Rates in Group Distributionally Robust Optimization via a Novel Notion of Sparsity

The minimax sample complexity of group distributionally robust optimization (GDRO) has been determined up to a log(K) factor, where K is the number of groups. In this work, we venture beyond the minimax perspective via a novel notion of sparsity that we call (λ, β)-sparsity. In short, this condition means that at any parameter θ, there is a set of at most β groups whose risks at θ are all at least λ larger than the risks of the other groups. To find an ϵ-optimal θ, we show via a novel algorithm and analysis that the ϵ-dependent term in the sample complexity can swap a linear dependence on K for a linear dependence on the potentially much smaller β. This improvement leverages recent progress in sleeping bandits, showing a fundamental connection between the two-player zerosum game optimization framework for GDRO and per-action regret bounds in sleeping bandits. We next show an adaptive algorithm which, up to logarithmic factors, obtains a sample complexity bound that adapts to the best (λ, β)-sparsity condition that holds. We also show how to obtain a dimension-free semi-adaptive sample complexity bound with a computationally efficient method. Finally, we demonstrate the practicality of the (λ, β)-sparsity condition and the improved sample efficiency of our algorithms on both synthetic and real-life datasets.