Conceptminimax regret lower bound / 遗憾下界
Minimax Regret Lower Bound
A theoretical lower bound within the minimax framework that characterizes the minimum cumulative regret any algorithm must incur in the worst case. It serves as a core tool for analyzing the optimality of algorithms in online learning and sequential decision-making settings.
Timeline (last 90 days)
Oct 9
The paper establishes a universal minimax regret lower bound Ω(√(Tγ_T / log T)) for non-constant continuous kernel functions on compact domains.
Unverified50%