The Expectation Maximization Algorithm
- The EM algorithm can also be used to learn a Bayes net.
- Calculate probabilities of unobserved variables, assuming
$h$.
- Calculate new $w_{ijk}$ to maximize $E[\ln P(D\,|\,h)]$ where $D$ now includes
both observed and (calculated probabilities of) unobserved variables.
- When the structure of the network is unknown we can use
algorithms that use greedy search to add/subtract edges and
nodes. This is an active area of research.
José M. Vidal
.
32 of 39