Bayesian estimation

likelihood

posterior (unnormalized)

Bayesian estimation • Goal is to choose the label (skin or ~skin) that maximizes the posterior – this is called Maximum A Posteriori (MAP) estimation

• Suppose the prior is uniform:  P(skin) = P(~skin) =

– in this case                                          , – maximizing the posterior is equivalent to maximizing the likelihood »                                                    if and only if – this is called Maximum Likelihood (ML) estimation