Access the full text.
Sign up today, get DeepDyve free for 14 days.
Yuri Boykov, O. Veksler, R. Zabih (2001)
Fast approximate energy minimization via graph cutsProceedings of the Seventh IEEE International Conference on Computer Vision, 1
G. Forney, F. Kschischang, B. Marcus, Selim Tuncel (2001)
Iterative Decoding of Tail-Biting Trellises and Connections with Symbolic Dynamics
J. Pearl (1988)
Probabilistic reasoning in intelligent systems
S. Aji, Gavin Horn, R. McEliece, Meina Xu (1998)
Iterative min-sum decoding of tail-biting codes1998 Information Theory Workshop (Cat. No.98EX131)
S. Aji, R. McEliece (2000)
The generalized distributive lawIEEE Trans. Inf. Theory, 46
F. Barahona (1982)
On the computational computational complexity of the Ising modelJournal of Physics A: Mathematical and General, 15
F. Barahona (1982)
On the computational complexity of Ising spin glass modelsJournal of Physics A, 15
(2002)
Topology-corrected belief revision in the ising model. Presented at Snowbird Learning workshop
J. Yedidia, W. Freeman, Yair Weiss (2000)
Generalized Belief Propagation
(1999)
Probablistic Networks and Expert Systems. Statistics for Engineering and Information Science
Yair Weiss, W. Freeman (2001)
On the optimality of solutions of the max-product belief-propagation algorithm in arbitrary graphsIEEE Trans. Inf. Theory, 47
S. Benedetto, G. Montorsi, D. Divsalar, F. Pollara (1996)
Soft-Output Decoding Algorithms in Iterative Decoding of Turbo Codes
A. Dawid (1992)
Applications of a general propagation algorithm for probabilistic expert systemsStatistics and Computing, 2
Richard Stanley (1986)
Enumerative Combinatorics
(1999)
Learning in graphical models
J. Ortega, W. Rheinboldt (2014)
Iterative solution of nonlinear equations in several variables
D Proof of Lemma
W. Freeman, E. Pasztor (1998)
Learning to Estimate Scenes from Images
J. Kleinberg, É. Tardos (1999)
Approximation algorithms for classification problems with pairwise relationships: metric labeling and Markov random fields40th Annual Symposium on Foundations of Computer Science (Cat. No.99CB37039)
M. Wainwright, T. Jaakkola, A. Willsky (2003)
Tree-based reparameterization framework for analysis of sum-product and related algorithmsIEEE Trans. Inf. Theory, 49
B. Frey, R. Kotter, A. Vardy (1998)
Skewness and pseudocodewords in iterative decodingProceedings. 1998 IEEE International Symposium on Information Theory (Cat. No.98CH36252)
G. Forney, R. Koetter, F. Kschischang, Alex Reznik (2001)
On the Effective Weights of Pseudocodewords for Codes Defined on Graphs with Cycles, 123
Wray Buntine (1994)
Learning with Graphical Models
Christian Schlegel, Lance Perez (2004)
Low-Density Parity-Check Codes
(2000)
Exact inference using the attenuated max-product algorithm. In Advanced mean field methods: Theory and Practice
J.M. Ortega, W.C. Rheinboldt (2000)
Classics in Applied Mathematics
Cycle codes correspond to low-density parity check codes (Gallager 1963) for which each bit is connected to exactly two parity checks
B. Bollobás (1998)
Modern Graph Theory
N. Wiberg (1996)
Codes and Decoding on General Graphs
C. Berge (1989)
Hypergraphs
R. McEliece, M. Yıldırım (2003)
Belief Propagation on Partially Ordered Sets
Yair Weiss (2000)
Correctness of Local Probability Propagation in Graphical Models with LoopsNeural Computation, 12
S.L. Lauritzen (1996)
Graphical models
A. Viterbi (1967)
Error bounds for convolutional codes and an asymptotically optimum decoding algorithmIEEE Trans. Inf. Theory, 13
C. White (1978)
Dynamic programming and stochastic controlProceedings of the IEEE, 66
B. Frey, R. Koetter, A. Vardy (2001)
Signal-space characterization of iterative decodingIEEE Trans. Inf. Theory, 47
This model corresponds to a collection of binary variables with attractive couplings between adjacent nodes
J. Forney (1973)
The Viterbi Algorithm
Alexander ScHRJJVER, A. Mayne (1989)
Geometric Algorithms and Combinatorial OptimizationJournal of the Operational Research Society
Y. Censor, S. Zenios (1997)
Parallel Optimization: Theory, Algorithms, and Applications
Y. Boykov, O. Veksler, R. Zabih (2001)
Fast approximate energy minimization via graph cutsIEEE. Trans. PAMI, 23
M. Wainwright (2002)
Stochastic processes on graphs with cycles: geometric and variational approaches
S. Verdú, V. Poor (1987)
Abstract dynamic programming models under commutativity conditionsSiam Journal on Control and Optimization, 25
J. Besag (1986)
On the Statistical Analysis of Dirty PicturesJournal of the royal statistical society series b-methodological, 48
45) does not directly specify τ 0 , but rather the quantities ϕ h (x; τ 0 h ) (which suffice to specify the parameterization of the distribution
D. Nilsson (1998)
An efficient algorithm for finding the M most probable configurationsin probabilistic expert systemsStatistics and Computing, 8
R. Cowell, A. Dawid, S. Lauritzen, D. Spiegelhalter (1999)
Probabilistic Networks and Expert Systems
M. Goemans, David Williamson (1995)
Improved approximation algorithms for maximum cut and satisfiability problems using semidefinite programmingJ. ACM, 42
L. Thomas (1983)
Optimization over Time. Dynamic Programming and Stochastic Control. Volume 1Journal of the Operational Research Society, 34
(2001)
Thesis Supervisor Accepted by.......................................................................
John Odentrantz (2000)
Markov Chains: Gibbs Fields, Monte Carlo Simulation, and QueuesTechnometrics, 42
J. Pearl (1991)
Probabilistic reasoning in intelligent systems - networks of plausible inference
G. Horn (1999)
Iterative decoding and pseudo-codewords
S.M. Aji, G. Horn, R.J. McEliece, M. Xu (1998)
Proc. IEEE Information Theory Workshop
H. Bodlaender (1993)
A Tourist Guide through TreewidthActa Cybern., 11
S. Geman, D. Geman (1984)
Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of ImagesIEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI-6
Finding the maximum a posteriori (MAP) assignment of a discrete-state distribution specified by a graphical model requires solving an integer program. The max-product algorithm, also known as the max-plus or min-sum algorithm, is an iterative method for (approximately) solving such a problem on graphs with cycles. We provide a novel perspective on the algorithm, which is based on the idea of reparameterizing the distribution in terms of so-called pseudo-max-marginals on nodes and edges of the graph. This viewpoint provides conceptual insight into the max-product algorithm in application to graphs with cycles. First, we prove the existence of max-product fixed points for positive distributions on arbitrary graphs. Next, we show that the approximate max-marginals computed by max-product are guaranteed to be consistent, in a suitable sense to be defined, over every tree of the graph. We then turn to characterizing the nature of the approximation to the MAP assignment computed by max-product. We generalize previous work by showing that for any graph, the max-product assignment satisfies a particular optimality condition with respect to any subgraph containing at most one cycle per connected component. We use this optimality condition to derive upper bounds on the difference between the log probability of the true MAP assignment, and the log probability of a max-product assignment. Finally, we consider extensions of the max-product algorithm that operate over higher-order cliques, and show how our reparameterization analysis extends in a natural manner.
Statistics and Computing – Springer Journals
Published: Oct 15, 2004
Read and print from thousands of top scholarly journals.
Already have an account? Log in
Bookmark this article. You can see your Bookmarks on your DeepDyve Library.
To save an article, log in first, or sign up for a DeepDyve account if you don’t already have one.
Copy and paste the desired citation format or use the link below to download a file formatted for EndNote
Access the full text.
Sign up today, get DeepDyve free for 14 days.
All DeepDyve websites use cookies to improve your online experience. They were placed on your computer when you launched this website. You can change your cookie settings through your browser.