In methods named for this man, plate notation simplifies repetitive parts of PGMs. This man’s namesake “regret” is bounded through (10[1])Thompson sampling, which applies a statement named for him after (-5[1])each step. In models named for him, (10[1])colliders and non-colliders (10[1])are (10[1])used to determine whether two vertices are d-separated. A data structure named for this man is used in his “variational” method, which is applied to models like latent Dirichlet (“dee-rick-LAY”) allocation. It’s not Schwarz, (10[1])but the Akaike (“ah-kah-EE-kay”) information (10[2])criterion (10[1])penalizes (10[1])complexity (10[1])less strongly (10[1])than (10[1])this man’s namesake model selection (10[1])criterion. MAP estimates are generated through this man’s namesake inference, (10[2])in (10[1])contrast (10[1])to MLE (10[1])estimates (10[1])generated through frequentist (10[3])inference. (10[1])For 10 points, what statistician (10[1])names a statement that relates prior and posterior conditional probabilities? ■END■