Quiz dossier / Deep Learning Book

18 - Confronting the Partition Function

Advanced mastery of Chapter 18, Confronting the Partition Function, covering every named section across book pages 605-630.

13 questionsEstimated 95 min0 flagged

Session status

Draft

Question 1

3 point(s)

Why the partition function is difficult

For $p(x)=\tilde p(x)/Z$ , why can maximum likelihood be intractable?

Z=\sum_x\tilde p(x)

or its integral may require enumerating exponentially many states and depends on the parameters

Z

is always exactly one before normalization

The unnormalized score cannot depend on parameters

Likelihood never requires probabilities