Week 4 Maximum Likelihood Flashcards
what is statistical interference
when you know the value of the bayes theorem equation and can test a hypothesis and fit a model to the data
what is the likelihood function equation
L(x I a) = II L(xi I a)
what is the maximum a posteriori
the most likely outcome of a
what are the MAP equation for a value a -give only one
MAP(a) = argmax P(a I x) = [δP(a I x) / δa] = 0 MAP(a) = argmax ln[P(a I x) = [δlnP(a I x) / δa] = 0
what is maximum likelihood estimation
MLE is the result of MAP when the prior probability is independent of value a
what do we use to determine the efficiency of MLE
variance
what is the RCF inequality equation
V(ahat) >= (1+δb/δa)^2 /
what happens to the RCF when the estimator is unbiased
it just becomes the minimum variance bound
what happens to the bias of an estimator as the number of trials increase
the bias vanishes
what two things can be said of the MLE
the logarithmic likelihood near the MLE can be approximated with a parabola so the likelihood is a Gaussian in the MLE region
The standard deviation on the MLE can be found by finding the value of the parameter for which the likelihood change is 1/2 wrt the MLE