Learning discrete decomposable graphical models via constraint optimization

View/ Open
File version
Accepted Manuscript (AM)
Author(s)
Janhunen, Tomi
Gebser, Martin
Rintanen, Jussi
Nyman, Henrik
Pensar, Johan
Corander, Jukka
Griffith University Author(s)
Year published
2015
Metadata
Show full item recordAbstract
Statistical model learning problems are traditionally solved using either heuristic greedy optimization or stochastic simulation, such as Markov chain Monte Carlo or simulated annealing. Recently, there has been an increasing interest in the use of combinatorial search methods, including those based on computational logic. Some of these methods are particularly attractive since they can also be successful in proving the global optimality of solutions, in contrast to stochastic algorithms that only guarantee optimality at the limit. Here we improve and generalize a recently introduced constraint-based method for learning ...
View more >Statistical model learning problems are traditionally solved using either heuristic greedy optimization or stochastic simulation, such as Markov chain Monte Carlo or simulated annealing. Recently, there has been an increasing interest in the use of combinatorial search methods, including those based on computational logic. Some of these methods are particularly attractive since they can also be successful in proving the global optimality of solutions, in contrast to stochastic algorithms that only guarantee optimality at the limit. Here we improve and generalize a recently introduced constraint-based method for learning undirected graphical models. The new method combines perfect elimination orderings with various strategies for solution pruning and offers a dramatic improvement both in terms of time and memory complexity. We also show that the method is capable of efficiently handling a more general class of models, called stratified/labeled graphical models, which have an astronomically larger model space.
View less >
View more >Statistical model learning problems are traditionally solved using either heuristic greedy optimization or stochastic simulation, such as Markov chain Monte Carlo or simulated annealing. Recently, there has been an increasing interest in the use of combinatorial search methods, including those based on computational logic. Some of these methods are particularly attractive since they can also be successful in proving the global optimality of solutions, in contrast to stochastic algorithms that only guarantee optimality at the limit. Here we improve and generalize a recently introduced constraint-based method for learning undirected graphical models. The new method combines perfect elimination orderings with various strategies for solution pruning and offers a dramatic improvement both in terms of time and memory complexity. We also show that the method is capable of efficiently handling a more general class of models, called stratified/labeled graphical models, which have an astronomically larger model space.
View less >
Journal Title
Statistics and Computing
Copyright Statement
© 2007 Springer US. This is an electronic version of an article published in Statistics and Computing, January 2017, Volume 27, Issue 1, pp 115–130. Statistics and Computing is available online at: http://link.springer.com/ with the open URL of your article.
Note
This publication has been entered into Griffith Research Online as an Advanced Online Version.
Subject
Computation Theory and Mathematics not elsewhere classified
Statistics
Computation Theory and Mathematics