Title :
Universality of the Local Marginal Polytope
Author :
Prusa, Daniel ; Werner, Tomas
Author_Institution :
Dept. of Cybern., Czech Tech. Univ., Prague, Czech Republic
Abstract :
We show that solving the LP relaxation of the min-sum labeling problem (also known as MAP inference problem in graphical models, discrete energy minimization, or valued constraint satisfaction) is not easier than solving any linear program. Precisely, every polytope is linear-time representable by a local marginal polytope and every LP can be reduced in linear time to a linear optimization (allowing infinite costs) over a local marginal polytope. The reduction can be done (though with a higher time complexity) even if the local marginal polytope is restricted to have a planar structure.
Keywords :
computational complexity; graph theory; linear programming; maximum likelihood estimation; LP relaxation; MAP inference problem; discrete energy minimization; graphical model; linear optimization; linear programming; local marginal polytope; maximum a priori estimation; min-sum labeling problem; time complexity; valued constraint satisfaction; Complexity theory; Encoding; Equations; Face; Minimization; Optimization; Vectors; Graphical model; Markov random field; discrete energy minimization; linear programming relaxation; local marginal polytope; valued constraint satisfaction;
Journal_Title :
Pattern Analysis and Machine Intelligence, IEEE Transactions on
DOI :
10.1109/TPAMI.2014.2353626