Tag
This paper proposes a Bayesian decision formulation for multiobjective optimization under uncertainties, emphasizing expected hypervolume maximization with Gaussian Processes and active learning strategies.
This paper formalizes LLM configuration evaluation as a cost-aware multi-objective bandit problem, proposing a hypervolume-based UCB algorithm for online configuration selection and a cost-aware gap elimination algorithm for Pareto identification, both with theoretical guarantees and empirical validation.
This paper introduces THV-UCB, an algorithm for multi-objective bandit problems with slate selection, and establishes gap-free and gap-dependent regret bounds for hypervolume regret.