2.5. Online.Bandit.Regret
Regret, gap, best arm
Main definitions
-
gap ν ais the gap of an actiona, i.e., the difference between the highest mean of the actions and the mean ofa. -
regret ν A t ωis the regret of a sequence of pullsA : ℕ → Ω → 𝓐at timetfor the reward kernelν : Kernel 𝓐 ℝand the outcomeω : Ω. -
bestArm νis an action with the highest mean.
Module LeanMachineLearning.Online.Bandit.Regret contains 18 exposed declarations.
-
Bandits.gap -
Bandits.gap_nonneg -
Bandits.gap_nonneg_of_le -
Bandits.gap_le_of_mem_Icc -
Bandits.regret -
Bandits.regret_eq_sum_gap -
Bandits.regret_nonneg -
Bandits.gap_eq_zero_of_regret_eq_zero -
Bandits.regret_eq_sum_pullCount_mul_gap -
Bandits.integral_regret_eq_sum_gap_mul_integral_pullCount -
Bandits.integral_regret_le_of_forall_integral_pullCount_le -
Bandits.bestArm -
Bandits.le_bestArm -
Bandits.gap_eq_bestArm_sub -
Bandits.gap_bestArm -
Bandits.integral_eq_of_gap_eq_zero -
Bandits.avg_mean_reward_tendsto_of_sublinear_regret -
Bandits.pullCount_rate_tendsto_of_sublinear_regret