6
$\begingroup$

I am trying to determine parameters in a bayesian network with two latent variables (in blue).

DAG (not enough reputation to post an image)

Every variable is discrete with 2-4 categories. The latent variables have 3 categories each.

I am trying to find an R package that will let me define the DAG and then learn the conditional probability parameters from data:

    B C D E F G H I J
    2 1 ? 3 2 ? 1 4 4
    2 2 ? 3 2 ? 1 3 2
    2 1 ? 3 2 ? 1 4 2
    3 3 ? 3 2 ? 1 4 2
    3 1 ? 3 2 ? 1 1 4
    3 3 ? 3 2 ? 1 4 1
    ...

I have tried working with:

  1. bnlearn: as far as I can tell the parameter estimation doesn't support latent variables
  2. catnet: I can't really tell what is going on here, but the results haven't worked out the way I expected them to when using the parameter estimation tools in this package.
  3. gRain: I tried working with this a lot, but couldn't sort out how to generate the CPT values when one or more of the variables in my DAG was latent.

I also tried working out how to use rJags for a long time but ended up giving up. The tutorials I found for parameter estimation all assumed either:

  1. There are no latent variables
  2. We were trying to do learning on the structure, rather than just determine conditional probabilities given a structure.

I'd really like it if I could find a solid R package for tackling this problem, though a java/scala solution would work as well. A tutorial/example problem would be even better.

I have a vague impression that expectation maximization techniques can be used and even started trying to write code for the problem myself by extending the cluster graph stuff written here bayes-scala but quickly realized that this was a bigger project than I initially thought.

Bayes-Scala
I worked with Bayes scala some more and found a video from a coursera class defining cluster graphs http //www.youtube.com/watch?v=xwD_B31sElc

I'll walk through here quickly what I think I learned. First off, the full DAG I am dealing with is (http //i.imgur.com/tnVPFGM.png) (still not enough reputation for images or more links)

Most of the language dealing with cluster graphs refers to factors in factor graphs, so I tried to translate this DAG into a factor graph (http //i.imgur.com/V54XdTV.png). I'm not certain how this works, but I think this factor graph is a valid transformation from the DAG.

Then a simple cluster graph I could make is called a bethe cluster graph (http //i.imgur.com/hHb7eQ7.png). The bethe cluster graph is guaranteed to satisfy the constraints of cluster graphs. Namely:

1. Family Preservation: For each factor $\Phi_k$ there must exist a cluster $C_i$ s.t $scope(\Phi_k) \subseteq C_i$

2. Running Intersection Property: for each pair of clusters $C_i,C_j$ and variable $X_i \in C_i \cap C_j$ there exists a unique path between $C_i,C_j$ such that all clusters along the path contain $X_i$

However, I think the bethe cluster graph loses the covariability conferred by the original DAG. So I tried to construct what I believe is a valid cluster graph that still maintains information about covariance between variables:
http //i.imgur.com/54gDJhl.png

I think its possible that I messed something up in this cluster graph, but I could use bayes scala to write this cluster graph (http //pastebin.com/82ErCSQT). After I do so the log likelihood's progress as one would imagine:

EM progress(iterNum, logLikelihood): 1, -2163.880428036244
EM progress(iterNum, logLikelihood): 2, -1817.7287344711604

and the marginal probabilities can be calculated. Unfortunately the marginal probabilities for the hidden nodes turn out to be flat after using GenericEMLearn:

marginal D: WrappedArray(0.25, 0.25, 0.25000000000000006, 0.25000000000000006)
marginal J: WrappedArray(0.25000000000000006, 0.25, 0.25, 0.25)

Additionally an odd error occurs where if I set evidence in a LoopyBP more than once I start getting NA values back:


val loopyBP = LoopyBP(skinGraph)
    loopyBP.calibrate()
    (1 to 44).foreach((sidx) => {
      val s1 = dataSet.samples(sidx)
      (1 to 10).foreach((x) => {      
        loopyBP.setEvidence(x,s1(x))
      })
      loopyBP.marginal(11).getValues.foreach((x) => print(x + " "))
      println()
    })
actual value was 0
0.0 0.0 0.0 1.0 
actual value was 0
NaN NaN NaN NaN 

This problem actually occurs for the sprinkler example as well, so I must be using setEvidence incorrectly somehow.

$\endgroup$
2
$\begingroup$

Regarding to "...Additionally an odd error occurs where if I set evidence in a LoopyBP more than once I start getting NA ..."

I think it's because you set the evidence in the same Bayesian Network iterating over all samples in a training set and then you end up with some factors having 0 probabilities for all factor values. How setEvidence() works for discrete factors is that it sets the evidence probability to 0 for all factor values incompatible with evidence. I will throw an error if setEvidence() is set for all factor values.

Why would you want to set the evidence in a single bayesian network over all samples?

Regarding to flat probabilities for hidden nodes, remember that EM does not guarantee to converge to a global maximum, it's quite important how you set the priors in the network. Please send me your code including training set and I will check it further.

On factor graph: Bayes-scala also supports inference on a factor graphs but only for continuous and hybrid bayesian networks using Expectation Propagation algorithm, the two cases I tested include kalman robot localisation and TrueSkill rating model.

And the last thing on sepsets including more than one variable. It's the current limitation of bayes-scala to allow for a single sepset variable.

$\endgroup$
  • $\begingroup$ The data is: pastebin.com/wktYyefF The DAG is: i.imgur.com/tnVPFGM.png I was iterating over samples to test if the network could be used to predict variables for which I did not set the evidence (I'd like to use this network to predict the H variable given values for all the other non-latent variables). $\endgroup$ – Thomas Luechtefeld Jun 8 '13 at 16:16
  • $\begingroup$ I wrote a method for doing predictions using a LoopyBP object here. I think ideally we don't have to copy the LoopyBP clustergraph, and instead just modify the message sending given evidence when calculating a marginal, but I haven't diven into the code enough to implement that yet. This function resolve the NA problem. $\endgroup$ – Thomas Luechtefeld Jun 8 '13 at 17:23
2
$\begingroup$

I'm the creator of bayes-scala toolbox you are referring to. Last year I implemented the EM in discrete bayesian network for learning from incomplete data (including latent variables), that looks like the use case you are asking about.

Some tutorial for a sprinkler bayesian network is here.

And, "Learning Dynamic Bayesian Networks with latent variables" is given here.

$\endgroup$
  • $\begingroup$ When I tried to translate my DAG into a cluster graph I first got a sepset not equal to 1 error because I handled the D,E,F region of the DAG incorrectly. After fixing that problem when I run GenericEMLearn I end up getting positive log likelihoods, and NA values for marginals. I do not fully understanding how to make a cluster graph out of a factor graph or DAG. I looked around for a resource on this and could not find one, do you have a suggestion for how to make a cluster graph from a DAG? The sprinkler example you gave doesn't fully define that transformation. $\endgroup$ – Thomas Luechtefeld Jun 7 '13 at 12:54
  • $\begingroup$ I will have a look during the weekend and try to learn your network with bayes-scala $\endgroup$ – Daniel Korzekwa Jun 7 '13 at 12:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.