If we have variables Xi, i=1,2,...n, and a variable Y where suppose each of the variables are having p states then for small p and small n producing CPT for X v/s Y is easier. When value of n and p increases suppose n=10 and p=2 then the number of instantiations of the X combinations should be at least (2^10)*2. The second 2 is for Y instantiations. So, this can become prohibitive for availability of processing the CPT. Can anyone give me some method to attack this problem?
Please look into this, if you have ever worked with R. http://rss.acs.unt.edu/Rdoc/library/Grappa/html/tab.html
Thanks Arun for your reply...I think you have not understood my problem completely... The soln you have stated will simply set up the CPT...I want a soln for large scale CPTs...You may call it as sparsity of CPTs, where number of instances of combinations is lower than the number of possible combinations of instantiations of variables...I hope the problem is clear now...
there is information how to deal with information type when filling in CPTs here: https://www.norsys.com/downloads/BBN%20Guidelines%20-%20Cain.pdf
- Badges
- Science topic