Learning bounded tree-width Bayesian networks using integer linear programming
2014 (English)In: Journal of machine learning research, ISSN 1532-4435, E-ISSN 1533-7928, Vol. 33, 751-759 p.Article in journal (Refereed) PublishedText
In many applications one wants to compute conditional probabilities given a Bayesian network. This inference problem is NP-hard in general but becomes tractable when the network has low tree-width. Since the inference problem is common in many application areas, we provide a practical algorithm for learning bounded tree-width Bayesian networks. We cast this problem as an integer linear program (ILP). The program can be solved by an anytime algorithm which provides upper bounds to assess the quality of the found solutions. A key component of our program is a novel integer linear formulation for bounding tree-width of a graph. Our tests clearly indicate that our approach works in practice, as our implementation was able to find an optimal or nearly optimal network for most of the data sets.
Place, publisher, year, edition, pages
Microtome Publishing , 2014. Vol. 33, 751-759 p.
Artificial intelligence, Forestry, Inference engines, Integer programming, Learning algorithms, Optimization, Trees (mathematics), Any-time algorithms, Application area, Conditional probabilities, Inference problem, Integer Linear Programming, Integer linear programs, Linear formulation, Optimal networks, Bayesian networks
Electrical Engineering, Electronic Engineering, Information Engineering
IdentifiersURN: urn:nbn:se:kth:diva-187580ScopusID: 2-s2.0-84955497631OAI: oai:DiVA.org:kth-187580DiVA: diva2:937144
22 April 2014 through 25 April 2014
FunderSwedish Research Council, 2013-4993
QC 201606152016-06-152016-05-252016-06-15Bibliographically approved