Welcome to BLIP!
BLIP stands for Bayesian network Learning and Inference Package
This webservice can learn huge Bayesian Networks from data, even with a very high number of variables.
Try it and see for yourself!
It has been developed by Mauro Scanagatta,
under the Imprecise Probability Group
The theory behind Blip has been published on:
• Learning Bayesian Networks with Thousands of Variables [NIPS 2015]
• Learning Treewidth-Bounded Bayesian Networks with Thousands of Variables [NIPS 2016] - (supplementary material)
The code for the bounded treewidth learning is available at: https://github.com/mauro-idsia/blip.
Feel free to use it as you wish, as long as you cite the original paper.
How can I use it?
Once you obtain a working accont here, you'll be able to submit a new dataset to evaluate.
Once the queue before you has been served your dataset will be analyzed, and you will obtain a full Bayesian Network.
Keep in mind that this kind of computation can be really cumbersome; we'll make our best, but under full usage some waiting time is to be expected.
Need additional help?