Scribd
Upload
41 views1 page

Linear Programming

Tree-reweighted belief propagation (TRBP) is a variant of belief propagation that finds the same solutions as linear programming (LP) relaxations, which are commonly used to approximate solutions to combinatorial optimization problems. While TRBP has similar time and space complexity to ordinary belief propagation, allowing it to be applied to large graphical models, its convergence properties and number of iterations needed for large problems is unknown. Experiments found that TRBP finds solutions significantly faster than solvers in CPLEX and can be used to find MAP configurations for large real-world problems in minutes, unlike the CPLEX solvers.

Uploaded by

Kshitij Agarwal
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
41 views1 page

Linear Programming

Tree-reweighted belief propagation (TRBP) is a variant of belief propagation that finds the same solutions as linear programming (LP) relaxations, which are commonly used to approximate solutions to combinatorial optimization problems. While TRBP has similar time and space complexity to ordinary belief propagation, allowing it to be applied to large graphical models, its convergence properties and number of iterations needed for large problems is unknown. Experiments found that TRBP finds solutions significantly faster than solvers in CPLEX and can be used to find MAP configurations for large real-world problems in minutes, unlike the CPLEX solvers.

Uploaded by

Kshitij Agarwal
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 1

c 


c    are a standard method for approximating combinatorial
optimization problems in computer science (Bertismas and Ttsitskikilis, 1997). They have been
used for approximating the MAP problem in a general graphical model by Santos (1991). More
recently, LP relaxations have been used for error-correcting codes (Feldman et al., 2003), and for
protein folding (Kingsford et al., 2005). LP relaxations have an advantage over other approximate
inference schemes in that they come with an optimality guarantee ± when the solution to the linear
program is integer, then the LP solution is guaranteed to give the global optimum of the posterior
probability.

Tree-reweighted belief propagation (TRBP) is a variant of belief propagation (BP) suggested by


Wainwright and colleagues (Wainwright et al., 2002), that has been shown to find the same solution
as LP relaxations. Each iteration of TRBP is similar in time and space complexity to that of ordinary
BP and hence it can be straightforwardly applied to very large graphical models. However, little is
known regarding the convergence properties of TRBP nor about the actual number of iterations
needed to solve large problems.

We find that TRBP almost always finds the solution significantly faster than all the solvers
in CPLEX and more importantly, TRBP can be applied to large scale problems for which the solvers
in CPLEX cannot be applied. Using TRBP we can find the MAP configurations in a matter of
minutes for a large range of real world problems.Y

You might also like