Learning: Likelihood with Missing Data4.2.2-LikelihoodwithMissingData.flv4.2.2-quiz14.2.2-quiz1.jpg22
vertLeftWhite1_2c_mid
The sum of two unimodal functions is generally bimodal4.2.2-quiz24.2.2-quiz2.jpg1,22
vertLeftWhite1_3c
The likelihood function doesn't distinguish between the true parameters and the other global optima, hence there is no way to know if a given set of parameters is the true one using the likelihood function. Furthermore, with a limited amount of data, the true parameters might not even be an optimum (local or global) of the likelihood function. This is why we may overfit with limited training data.4.2.2-quiz34.2.2-quiz3.jpg22
vertLeftWhite1_3c
There is now an active path between theta_(Y | X) and theta_X through the v-structure with Y as the observed child.