BayesFusion Support Forum

Thanks so much for your time!

Thanks, It's probably something I'm doing (or not doing).

I seem to be able to set evidence with
node->Value()->SetEvidence(double)

I set the evidence for all but one variable, set that variable as the target
network->SetTarget(int nodeID)

and run inference on the network with UpdateBeliefs but it never returns after several minutes (half hour or more ...

Are there any references for setting evidence and inference with networks composed of continuous variables? I haven't found anything in my search of the forum and reference documentation.

Thanks as always,
Greg Johnson

When I invoke ToDAG after DSL_pc::Learn DSL_pattern::ToDAG returns false but I get the following output from DSL_pattern::Print which shows no undirected arcs.

None: 0
Undirected: 1
Directed: 2
0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2
0 0 2 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 ...

Perhaps I am confused about how to use the pattern, dataset and network. After a dataset has been created, I have:

if (!pattern.ToDAG(d, result))
cout << "\tDSL_pattern::ToDAG returned false!"<< endl;


pattern.Print();

cout << "Starting learning... ";

if (error_code = pc.Learn(d,pattern ...

Okay, when I use a different dataset, or a concatenation of several datasets, DSL_pattern::Print produces the following output:

None: 0
Undirected: 1
Directed: 2

And DSL_pc::Learn doesn't produce a valid network although no errors are produced/detected.

I've got the columns copied to continuous variables, Now I get a "pc: constant not allowed" error. Do I have a column with too little variation? I've tried adjusting the pc significance value with no success.

Thanks,
Greg

pc.h is 720 bytes and dated 4/16/09 01:43:17 PM

I'm afraid I don't understand what you mean about copying columns or the dataset to ensure all columns are continuous.

Thanks for the quick response!

pc::Learn returns 1 not -1 as expected. However, the error message indicates that there is a mix of continuous and discrete variables. Is there a way to force all variables to be continuous? I have edited the dataset so all data elements have at least one decimal ...

Hi,

I am trying to learn a network with continuous variables by modifying code that worked with the previous version of SmiLearn.

The dataset has 30 variables and approx 4800 records. I have run tests where all data are real numbers with no change in the results (the zero values are integers in ...

Yes, I was using DSL_OKAY!!!

Thanks for the quick response.

I see now that these have been removed from the API.

DSL_dataset::ReadFile seems to fail on files with variable states which include an underscore. Is there another delimiter I can use? Currently I am using a space to separate the variable state names in the datafile.

Thanks.

Hi,

I'm trying to port my Visual Studio 6.0 code to Linux and am getting the following errors:

LearningTest.cpp: In function ‘int main(int, const char**)’:
LearningTest.cpp:9: error: ‘DSL_textParser’ was not declared in this scope
LearningTest.cpp:9: error: expected `;' before ‘parserâ ...

shooltz wrote:Unfortunately, we don't have the access to Altix hardware.

I believe I can get you an ID on our system temporarily to compile the libraries. Would you be able to do this for me and anyone else working on this platform?