difference between expectation maximisation and hypergeometric distribution for motif finding

Question

[kdc15]

Would someone be able to explain what are the differences in these two algorithms for motif finding? I am having trouble fully understanding what are the main differences, and which is a better approach to use.