|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
n |
Cope with missing values
|
|
|
|
n |
Hard to do in
closed form in Bayesian framework
|
|
|
|
n |
Approximations
|
|
|
n |
Multiple queries
|
|
|
|
n |
Currently
myopic: figuring out the single next best query
|
|
n |
Active learning
for network structure
|
|
|
|
n |
Need to
maintain distribution over structures: expensive
|
|
|
|
n |
Attempt to
obtain closed form solution for fixed ordering
|
|
|
|
n |
Then do MCMC
over orderings
|
|