CS242: Probabilistic Graphical Models Lecture 1B: Factor Graphs, Inference, & Variable Elimination

Total Page:16

File Type:pdf, Size:1020Kb

CS242: Probabilistic Graphical Models Lecture 1B: Factor Graphs, Inference, & Variable Elimination CS242: Probabilistic Graphical Models Lecture 1B: Factor Graphs, Inference, & Variable Elimination Professor Erik Sudderth Brown University Computer Science September 15, 2016 Some figures and materials courtesy David Barber, Bayesian Reasoning and Machine Learning http://www.cs.ucl.ac.uk/staff/d.barber/brml/ CS242: Lecture 1B Outline Ø Factor graphs Ø Loss Functions and Marginal Distributions Ø Variable Elimination Algorithms Factor Graphs set of hyperedges linking subsets of nodes f F ✓ V set of N nodes or vertices, 1, 2,...,N V 1 { } p(x)= f (xf ) Z = (x ) Z f f f x f Y2F X Y2F Z>0 normalization constant (partition function) (x ) 0 arbitrary non-negative potential function f f ≥ • In a hypergraph, the hyperedges link arbitrary subsets of nodes (not just pairs) • Visualize by a bipartite graph, with square (usually black) nodes for hyperedges • Motivation: factorization key to computation • Motivation: avoid ambiguities of undirected graphical models Factor Graphs & Factorization For a given undirected graph, there exist distributions with equivalent Markov properties, but different factorizations and different inference/learning complexities. 1 p(x)= (x ) Z f f f Y2F Undirected Pairwise (edge) Maximal Clique An Alternative Graphical Model Potentials Potentials Factorization 1 1 Recommendation: p(x)= (x ,x ) p(x)= (x ) Use undirected graphs Z st s t Z c c (s,t) c for pairwise MRFs, Y2E Y2C and factor graphs for higher-order potentials 6 K. Yamaguchi, T. Hazan, D. McAllester, R. Urtasun > 2pixels > 3pixels > 4pixels > 5pixels Non-Occ All Non-Occ All Non-Occ All Non-Occ All GC+occ [36] 39.76 % 40.97 % 33.50 % 34.74 % 29.86 % 31.10 % 27.39 % 28.61 % OCV-BM [37] 27.59 % 28.97 % 25.39 % 26.72 % 24.06 % 25.32 % 22.94 % 24.14 % CostFilter [38] 25.85 % 27.05 % 19.96 % 21.05 % 17.12 % 18.10 % 15.51 % 16.40 % GCS [26] 18.99 % 20.30 % 13.37 % 14.54 % 10.40 % 11.44 % 8.63 % 9.55 % GCSF [39] 20.75 % 22.69 % 13.02 % 14.77 % 9.48 % 11.02 % 7.48 % 8.84 % SDM [27] 15.29 % 16.65 % 10.98 % 12.19 % 8.81 % 9.87 % 7.44 % 8.39 % ELAS [30] 10.95 % 12.82 % 8.24 % 9.95 % 6.72 % 8.22 % 5.64 % 6.94 % OCV-SGBM [25] 10.58 % 12.20 % 7.64 % 9.13 % 6.04 % 7.40 % 5.04 % 6.25 % ITGV [40] 8.86 % 10.20 % 6.31 % 7.40 % 5.06 % 5.97 % 4.26 % 5.01 % Ours 6.25 % 7.78 % 4.13 % 5.45 % 3.18 % 4.32 % 2.66 % 3.66 % Example: Nearest-NeighborTable 2. Comparison Spatial with the state MRF of the art on the test set of KITTI [2] Stereo Vision: Yamaguchi et al., ECCV 2012 2 Diverse M-Best Solutions in Markov Random Fields Fig.input 4.ObjectKITTI MAP Segmentation: examples. mode (Left) Batra input Original.et al., MAP (Middle) ECCV mode 2012 Disparity. (Right) Disparity errors. Compatibility potential: We introduce an additional potential which ensures that the discrete occlusion labels match well the disparity observations. We do so by penalizing occlusion boundaries that are not supported by the data ˆ ˆ occ λimp if p Bij : di(p, yfront) < dj(p, yback) • Observed nodes: Features of 2D φimageij (yfront (intensity,, yback)= color, texture,9 2…) Fig. 1: Examples of category-level segmentation(0 results otherwise on test images from VOC 2010. • Hidden nodes: Property of 3DFor world each image, (depth, “mode” above motion, is the best ofobject 10 solutions category, obtained with Div…)MBest. Input MAPneg 2nd MAP 2nd Mode Input MAP Mode We also define φij (yi) to be a function which penalizes negative disparities ˆ neg λimp if minp Bij di(p, yi) < 0 φij (yi)= 2 (0 otherwise Fig. 2: (Left) Interactive segmentation setup; (Right) Pose tracking. “Mode” above refersWe to impose the solution a obtained regularization with DivMBest on. the type of occlusion boundary, where we prefer forsimpler a diverse explanations set of highly probable (i.e., solutions coplanar instead is from preferable our existing models. than hinge which is more desir- Even if the MAP solution alone is of poor quality, a diverse set of highly probably able than occlusion). We encode this preference by defining λocc > λhinge > 0. hypotheses might still enable accurate predictions. Note that in contrast to the M-BestWe thus MAP define problem [9 our, 22, 38 computability] that involves finding potential the top M most probable solutions under a probabilistic model, our approach emphasizes diversity.We seek to produce highly probable solutions thatneg are qualitativelyneg di↵erent fromocc the MAP and from each other. Thisλocc is+ anφij important(yi)+ distinctionφij (yj because)+φij the(yi, yj )ifoij = lo literal definition of M-best MAP is not expectedneg to work wellneg in practice.occ In a λocc + φ (yi)+φ (yj )+φij (yj , yi)ifoij = ro largebdy2 state-space problem (e.g.282,000,000) anyij reasonable settingij of M (10 50) φij (oij, yi, yj)= neg neg − 1 would return solutions nearly> identicalλhinge to+ theφij MAP.(yi Ideally,)+φij we( wouldyj )+ likeB to p B ∆dij if oij = hi > | ij | 2 ij find the modes of the distribution<> learntneg by our probabilisticneg model.1 Figs. 1, 2 φ (yi)+φ (yj )+ ∆di,j if oij = co show examples of these diverse solutionsij extractedij for di↵erent tasks.S S p SiPSj | i[ j | 2 [ Overview. In this paper, we introduce the Diverse M-Best problem, which > P involves finding a diverseˆ set of:> highly probableˆ solutions2 under a discrete prob- abilisticwith model.∆di,j Our=( formulationdi(p, yi assumes) dj access(p, y toj a)) dissimilarity. function ∆( , ) measuring the di↵erence between two− solutions. We present an integer program-· · ming formulation for the Diverse M-Best problem, study its Lagrangian relax- ation.Junction We show that Feasibility: this relaxation hasFollowing an interesting work interpretation on occlusion as the ∆- boundary reasoning [15, augmented16], we energy utilize minimization higher problem, order which potentials minimizes a to linear encode combination whether a junction of three of the energy and similarity to previous solutions, thereby producing solutions with low energy and high diversity. For simplicity we sometimes will refer to these solutions as modes, although they are not guaranteed to be actual modes of the Gibbs distribution. Higher-Order MRF Potentials Higher-order models in Computer Vision Kohli & Rother 2012 7 Figure 1.2: Some qualitative results. Please view in colour. First Row: Original Image. Second Row: • Observed nodes: Features of 2D image (intensity,Unary likelihood color, labeling from TextonBoost texture, [34]. Third Row: …) Result obtained using a pairwise contrast preserving smoothness potential as described in [34]. Fourth Row: Result obtained using the Pn Potts model • Hidden nodes: Property of 3D world (depth, motion,potential [15]. Fifth Row: object Results using the Robustcategory, Pn model potential (1.7) with…) truncation parameter Q = 0.1 c , where c is equal to the size of the super-pixel over which the Robust Pn higher-order potential | | | | is defined. Sixth Row: Hand labeled segmentations. The ground truth segmentation are not perfect and many pixels (marked black) are unlabelled. Observe that the Robust Pn model gives best results. For instance, the leg of the sheep and bird have been accurately labeled which was missing in the other results. Unlike the standard Pn Potts model, this potential function gives rise to a cost that is a linear truncated function of the number of inconsistent variables (see figure 1.1). This enables the robust potential to allow Low Density Parity Check (LDPC) Codes Parity Check Factors Ø Each variable node is binary, so xs 0, 1 Ø Parity check factors2 { } equal 1 if the sum of the connected bits is even, 0 if the sum is odd (invalid codewords Evidence (observation) Factors are excluded) Ø Unary evidence factors Data equal probability that Bits each bit is a 0 or 1, given data. Assumes independent “noise” on Parity each bit. Bits Inference: 0 5 15 \ Figure 3: Belief propagation decoding of a (3,6) regular LDPC code, given a binary input message (top half) corrupted by noise with error probability ϵ =0.08 (as set up in [5]). We show the received codeword including parity bits (zero iterations, left), the bits that maximize the posterior marginals after five and fifteen BP iterations, and the final, correctly decoded signal after convergence to a loopy BP fixed point (the Stata center, home of the MIT Computer Science & Artificial Intelligence Laboratory). example systematic message encoding [5], in which the first half of the transmitted bits are the uncoded message of interest. To aid visualization, we choose this message to be a simple binary image, but the LDPC code does not model or exploit this spatial structure. To decode a noise–corrupted message via loopy BP, we pass messages in a parallel fashion. Each iteration begins by sending new messages from each variable node to all neighboring factor nodes, and then updates the outgoing messages from each parity check. As illustrated in Fig. 3, this allows information from un–violated parity checks to propagate throughout the message, gradually correcting noisy bits. These BP message updates continue until the thresholded marginal probabilities define a valid codeword, or until some maximum number of iterations is reached. In practice, BP decoding typically fails due to oscillations in the message update dynamics, rather than convergence to an incorrect codeword [5]. The outstanding empirical performance of message–passing decoders has inspired extensive research on alternatives to regular LDPC codes [5]. One widely used trick removes all cycles of length four from the factor graph, so that no two parity checks share a common pair of message bits.
Recommended publications
  • Bayesian Network - Wikipedia
    2019/9/16 Bayesian network - Wikipedia Bayesian network A Bayesian network, Bayes network, belief network, decision network, Bayes(ian) model or probabilistic directed acyclic graphical model is a probabilistic graphical model (a type of statistical model) that represents a set of variables and their conditional dependencies via a directed acyclic graph (DAG). Bayesian networks are ideal for taking an event that occurred and predicting the likelihood that any one of several possible known causes was the contributing factor. For example, a Bayesian network A simple Bayesian network. Rain could represent the probabilistic relationships between diseases and influences whether the sprinkler symptoms. Given symptoms, the network can be used to compute the is activated, and both rain and probabilities of the presence of various diseases. the sprinkler influence whether the grass is wet. Efficient algorithms can perform inference and learning in Bayesian networks. Bayesian networks that model sequences of variables (e.g. speech signals or protein sequences) are called dynamic Bayesian networks. Generalizations of Bayesian networks that can represent and solve decision problems under uncertainty are called influence diagrams. Contents Graphical model Example Inference and learning Inferring unobserved variables Parameter learning Structure learning Statistical introduction Introductory examples Restrictions on priors Definitions and concepts Factorization definition Local Markov property Developing Bayesian networks Markov blanket Causal networks Inference complexity and approximation algorithms Software History See also Notes References https://en.wikipedia.org/wiki/Bayesian_network 1/14 2019/9/16 Bayesian network - Wikipedia Further reading External links Graphical model Formally, Bayesian networks are DAGs whose nodes represent variables in the Bayesian sense: they may be observable quantities, latent variables, unknown parameters or hypotheses.
    [Show full text]
  • An Introduction to Factor Graphs
    An Introduction to Factor Graphs Hans-Andrea Loeliger MLSB 2008, Copenhagen 1 Definition A factor graph represents the factorization of a function of several variables. We use Forney-style factor graphs (Forney, 2001). Example: f(x1, x2, x3, x4, x5) = fA(x1, x2, x3) · fB(x3, x4, x5) · fC(x4). fA fB x1 x3 x5 x2 x4 fC Rules: • A node for every factor. • An edge or half-edge for every variable. • Node g is connected to edge x iff variable x appears in factor g. (What if some variable appears in more than 2 factors?) 2 Example: Markov Chain pXYZ(x, y, z) = pX(x) pY |X(y|x) pZ|Y (z|y). X Y Z pX pY |X pZ|Y We will often use capital letters for the variables. (Why?) Further examples will come later. 3 Message Passing Algorithms operate by passing messages along the edges of a factor graph: - - - 6 6 ? ? - - - - - ... 6 6 ? ? 6 6 ? ? 4 A main point of factor graphs (and similar graphical notations): A Unified View of Historically Different Things Statistical physics: - Markov random fields (Ising 1925) Signal processing: - linear state-space models and Kalman filtering: Kalman 1960. - recursive least-squares adaptive filters - Hidden Markov models: Baum et al. 1966. - unification: Levy et al. 1996. Error correcting codes: - Low-density parity check codes: Gallager 1962; Tanner 1981; MacKay 1996; Luby et al. 1998. - Convolutional codes and Viterbi decoding: Forney 1973. - Turbo codes: Berrou et al. 1993. Machine learning, statistics: - Bayesian networks: Pearl 1988; Shachter 1988; Lauritzen and Spiegelhalter 1988; Shafer and Shenoy 1990. 5 Other Notation Systems for Graphical Models Example: p(u, w, x, y, z) = p(u)p(w)p(x|u, w)p(y|x)p(z|x).
    [Show full text]
  • Exact and Approximate Inference in Graphical Models
    Exact and approximate inference in graphical models: variable elimination and beyond Nathalie Dubois Peyrard, Simon de Givry, Alain Franc, Stephane Robin, Regis Sabbadin, Thomas Schiex, Matthieu Vignes To cite this version: Nathalie Dubois Peyrard, Simon de Givry, Alain Franc, Stephane Robin, Regis Sabbadin, et al.. Exact and approximate inference in graphical models: variable elimination and beyond. 2015. hal-01197655 HAL Id: hal-01197655 https://hal.archives-ouvertes.fr/hal-01197655 Preprint submitted on 5 Jun 2020 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Exact and approximate inference in graphical models: variable elimination and beyond Nathalie Peyrarda, Simon de Givrya, Alain Francb, Stephane´ Robinc;d, Regis´ Sabbadina, Thomas Schiexa, Matthieu Vignesa a INRA UR 875, Unite´ de Mathematiques´ et Informatique Appliquees,´ Chemin de Borde Rouge, 31326 Castanet-Tolosan, France b INRA UMR 1202, Biodiversite,´ Genes` et Communautes,´ 69, route d’Arcachon, Pierroton, 33612 Cestas Cedex, France c AgroParisTech, UMR 518 MIA, Paris 5e, France d INRA, UM R518 MIA, Paris 5e, France June 30, 2015 Abstract Probabilistic graphical models offer a powerful framework to account for the dependence structure between variables, which can be represented as a graph.
    [Show full text]
  • Conditional Random Fields
    Conditional Random Fields Grant “Tractable Inference” Van Horn and Milan “No Marginal” Cvitkovic Recap: Discriminative vs. Generative Models Suppose we have a dataset drawn from Generative models try to learn Hard problem; always requires assumptions Captures all the nuance of the data distribution (modulo our assumptions) Discriminative models try to learn Simpler problem; often all we care about Typically doesn’t let us make useful modeling assumptions about Recap: Graphical Models One way to make generative modeling tractable is to make simplifying assumptions about which variables affect which other variables. These assumptions can often be represented graphically, leading to a whole zoo of models known collectively as Graphical Models. Recap: Graphical Models One zoo-member is the Bayesian Network, which you’ve met before: Naive Bayes HMM The vertices in a Bayes net are random variables. An edge from A to B means roughly: “A causes B” Or more formally: “B is independent of all other vertices when conditioned on its parents”. Figures from Sutton and McCallum, ‘12 Recap: Graphical Models Another is the Markov Random Field (MRF), an undirected graphical model Again vertices are random variables. Edges show which variables depend on each other: Thm: This implies that Where the are called “factors”. They measure how compatible an assignment of subset of random variables are. Figure from Wikipedia Recap: Graphical Models Then there are Factor Graphs, which are just MRFs with bonus vertices to remove any ambiguity about exactly how the MRF factors An MRF An MRF A factor graph for this MRF Another factor graph for this MRF Circles and squares are both vertices; circles are random variables, squares are factors.
    [Show full text]
  • The Core in Random Hypergraphs and Local Weak Convergence 3
    THE CORE IN RANDOM HYPERGRAPHS AND LOCAL WEAK CONVERGENCE KATHRIN SKUBCH∗ Abstract. The degree of a vertex in a hypergraph is defined as the number of edges incident to it. In this paper we study the k-core, defined as the max- imal induced subhypergraph of minimum degree k, of the random r-uniform r−1 hypergraph Hr(n,p) for r ≥ 3. We consider the case k ≥ 2 and p = d/n for which asymptotically every vertex has fixed average degree d/(r − 1)! > 0. We derive a multi-type branching process that describes the local structure of the k-core together with the mantle, i.e. the vertices outside the k-core. This paper provides a generalization of prior work on the k-core problem in random graphs [A. Coja-Oghlan, O. Cooley, M. Kang, K. Skubch: How does the core sit inside the mantle?, Random Structures & Algorithms 51 (2017) 459–482]. Mathematics Subject Classification: 05C80, 05C65 (primary), 05C15 (sec- ondary) Key words: Random hypergraphs, core, branching process, Warning Prop- agation 1. Introduction r−1 Let r ≥ 3 and let d > 0 be a fixed constant. We denote by H = Hr(n,d/n ) the random r-uniform hypergraph on the vertex set [n] = {1,...,n} in which any subset a of [n] of cardinality r forms an edge with probability p = d/nr−1 inde- pendently. We say that the random hypergraph H enjoys a property with high probability (‘w.h.p.’) if the probability of this event tends to 1 as n tends to infinity. 1.1.
    [Show full text]
  • 4 Factor Graphs and Comparing Graphical Model Types
    Massachusetts Institute of Technology Department of Electrical Engineering and Computer Science 6.438 Algorithms for Inference Fall 2014 4 Factor graphs and Comparing Graphical Model Types We now introduce a third type of graphical model. Beforehand, let us summarize some key perspectives on our first two. First, for some ordering of variables, we can write any probability distribution as px(x1; : : : ; xn) = p (x1)p j (x2; x1) p j ;:::; (xn x1; : : : ; xn−1); x1 x2 x1 ··· xn x1 xn−1 j which can be expressed as a fully connected directed graphical model. When the conditional distributions involved do not depend on all the indicated conditioning variables, then some of the edges in the directed graphical model can be removed. This reduces the complexity of inference, since the associated conditional probability tables have a more compact description. By contrast, undirected graphical models express distributions of the form 1 Y px(x1; : : : ; xn) = Φc(xc); Z c2C where the potential functions Φc are non-negative and C is the set of maximal cliques on an undirected graph G. Remember that the Hammersley-Clifford theorem says that any distribution that factors in this way satisfies the Markov property on the graph and conversely that if p is strictly positive and satisfies the Markov property for G then it factors as above. Evidently, any distribution can be expressed using fully connected undirected graphical model, since this corresponds to a single potential involving all the variables—i.e., a joint probability distribution. Undirected graphical models efficiently represent conditional independencies in the distribution of interest, which are expressed by the removal of edges, and which similarly reduces the complexity of inference.
    [Show full text]
  • Symbolic Variable Elimination for Discrete and Continuous Graphical Models
    Proceedings of the Twenty-Sixth AAAI Conference on Artificial Intelligence Symbolic Variable Elimination for Discrete and Continuous Graphical Models Scott Sanner Ehsan Abbasnejad NICTA & ANU ANU & NICTA Canberra, Australia Canberra, Australia [email protected] [email protected] Abstract Probabilistic reasoning in the real-world often requires inference in continuous variable graphical models, yet there are few methods for exact, closed-form inference when joint distributions are non-Gaussian. To address this inferential deficit, we introduce SVE – a symbolic Figure 1: The robot localization graphical model and all condi- extension of the well-known variable elimination al- tional probabilities. gorithm to perform exact inference in an expressive class of mixed discrete and continuous variable graphi- and xi are the observed measurements of distance (e.g., us- cal models whose conditional probability functions can ing a laser range finder). The prior distribution P (d) is uni- be well-approximated as oblique piecewise polynomi- form U(d; 0; 10) while the observation model P (xijd) is a als with bounded support. Using this representation, we mixture of three distributions: (red) is a truncated Gaussian ex- show that we can compute all of the SVE operations representing noisy measurements x of the actual distance d actly and in closed-form, which crucially includes defi- i nite integration w.r.t. multivariate piecewise polynomial within the sensor range of [0; 10]; (green) is a uniform noise functions. To aid in the efficient computation and com- model U(d; 0; 10) representing random anomalies leading pact representation of this solution, we use an extended to any measurement; and (blue) is a triangular distribution algebraic decision diagram (XADD) data structure that peaking at the maximum measurement distance (e.g., caused supports all SVE operations.
    [Show full text]
  • An Introduction to Conditional Random Fields
    Foundations and Trends R in sample Vol. xx, No xx (xxxx) 1{87 c xxxx xxxxxxxxx DOI: xxxxxx An Introduction to Conditional Random Fields Charles Sutton1 and Andrew McCallum2 1 EdinburghEH8 9AB, UK, [email protected] 2 Amherst, MA01003, USA, [email protected] Abstract Often we wish to predict a large number of variables that depend on each other as well as on other observed variables. Structured predic- tion methods are essentially a combination of classification and graph- ical modeling, combining the ability of graphical models to compactly model multivariate data with the ability of classification methods to perform prediction using large sets of input features. This tutorial de- scribes conditional random fields, a popular probabilistic method for structured prediction. CRFs have seen wide application in natural lan- guage processing, computer vision, and bioinformatics. We describe methods for inference and parameter estimation for CRFs, including practical issues for implementing large scale CRFs. We do not assume previous knowledge of graphical modeling, so this tutorial is intended to be useful to practitioners in a wide variety of fields. Contents 1 Introduction 1 2 Modeling 5 2.1 Graphical Modeling 6 2.2 Generative versus Discriminative Models 10 2.3 Linear-chain CRFs 18 2.4 General CRFs 21 2.5 Applications of CRFs 23 2.6 Feature Engineering 24 2.7 Notes on Terminology 26 3 Inference 27 3.1 Linear-Chain CRFs 28 3.2 Inference in Graphical Models 32 3.3 Implementation Concerns 40 4 Parameter Estimation 43 i ii Contents 4.1 Maximum Likelihood 44 4.2 Stochastic Gradient Methods 52 4.3 Parallelism 54 4.4 Approximate Training 54 4.5 Implementation Concerns 61 5 Related Work and Future Directions 63 5.1 Related Work 63 5.2 Frontier Areas 70 1 Introduction Fundamental to many applications is the ability to predict multiple variables that depend on each other.
    [Show full text]
  • Markov Random Field Modeling, Inference & Learning in Computer
    Markov Random Field Modeling, Inference & Learning in Computer Vision & Image Understanding: A Survey Chaohui Wang, Nikos Komodakis, Nikos Paragios To cite this version: Chaohui Wang, Nikos Komodakis, Nikos Paragios. Markov Random Field Modeling, Inference & Learning in Computer Vision & Image Understanding: A Survey. Computer Vision and Image Un- derstanding, Elsevier, 2013, 117 (11), pp.1610-1627. 10.1016/j.cviu.2013.07.004. hal-00858390v2 HAL Id: hal-00858390 https://hal.archives-ouvertes.fr/hal-00858390v2 Submitted on 6 Sep 2013 HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. Markov Random Field Modeling, Inference & Learning in Computer Vision & Image Understanding: A Survey Chaohui Wanga,b, Nikos Komodakisa,c, Nikos Paragiosa,d aCenter for Visual Computing, Ecole Centrale Paris, Grande Voie des Vignes, Chatenay-Malabry, France bPerceiving Systems Department, Max Planck Institute for Intelligent Systems, T¨ubingen, Germany cLIGM laboratory, University Paris-East & Ecole des Ponts Paris-Tech, Marne-la-Vall´ee, France dGALEN Group, INRIA Saclay - Ile de France, Orsay, France Abstract In this paper, we present a comprehensive survey of Markov Random Fields (MRFs) in computer vision and image understanding, with respect to the modeling, the inference and the learning.
    [Show full text]
  • Factor Graphs and Message Passing
    Factor Graphs and message passing Carl Edward Rasmussen October 28th, 2016 Carl Edward Rasmussen Factor Graphs and message passing October 28th, 2016 1 / 13 Key concepts • Factor graphs are a class of graphical model • A factor graph represents the product structure of a function, and contains factor nodes and variable nodes • We can compute marginals and conditionals efficiently by passing messages on the factor graph, this is called the sum-product algorithm (a.k.a. belief propagation or factor-graph propagation) • We can apply this to the True Skill graph • But certain messages need to be approximated • One approximation method based on moment matching is called Expectation Propagation (EP) Carl Edward Rasmussen Factor Graphs and message passing October 28th, 2016 2 / 13 Factor Graphs Factor graphs are a type of probabilistic graphical model (others are directed graphs, a.k.a. Bayesian networks, and undirected graphs, a.k.a. Markov networks). Factor graphs allow to represent the product structure of a function. Example: consider the factorising probability distribution: p(v, w, x, y, z) = f1(v, w)f2(w, x)f3(x, y)f4(x, z) A factor graph is a bipartite graph with two types of nodes: • Factor node: Variable node: • Edges represent the dependency of factors on variables. Carl Edward Rasmussen Factor Graphs and message passing October 28th, 2016 3 / 13 Factor Graphs p(v, w, x, y, z) = f1(v, w)f2(w, x)f3(x, y)f4(x, z) • What are the marginal distributions of the individual variables? • What is p(w)? • How do we compute conditional distributions, e.g.
    [Show full text]
  • Sum-Product Inference for Factor Graphs, Learning Directed Graphical Models
    Probabilistic Graphical Models Brown University CSCI 2950-P, Spring 2013 Prof. Erik Sudderth Lecture 6: Sum-Product Inference for Factor Graphs, Learning Directed Graphical Models Some figures courtesy Michael Jordan’s draft textbook, An Introduction to Probabilistic Graphical Models 50 CHAPTER 2. NONPARAMETRIC AND GRAPHICAL MODELS Pairwise Markov Random Fields x1 1 x2 x1 x2 x1 x2 p(x)= (x ,x ) (x ) Z st s t s s (s,t) s Y2E Y2V • Simple xparameterization,3 but xstill3 x3 expressive and widely used in practice • Guaranteed Markov with respect to graph x4 x5 x4 x5 x4 x5 (a) (b) (c) Figure 2.4. Three graphical representations of a distribution over five random variables (see [175]). (a) Directed graph G depicting a causal, generative process. (b) Factor graph expressing the factoriza- tion underlying G. (c) A “moralized” undirected graph capturing the Markov structure of G. set of undirected edges (s,t) linking pairs of nodes For example,E in the factor graph of Fig. 2.5(c), there are 5 variable nodes, and the joint set of N nodes or vertices, 1, 2,...,N distributionV has one potential for each of the 3{ hyperedges: } normalizationp(x) ψ (x constant,x ,x ) ψ (partition(x ,x ,x function)) ψ (x ,x ) Z ∝ 123 1 2 3 234 2 3 4 35 3 5 Often, these potentials can be interpreted as local dependencies or constraints. Note, however, that ψf (xf ) does not typically correspond to the marginal distribution pf (xf ), due to interactions with the graph’s other potentials. In many applications, factor graphs are used to impose structure on an exponential family of densities.
    [Show full text]
  • Classical Probabilistic Models and Conditional Random Fields
    Classical Probabilistic Models and Conditional Random Fields Roman Klinger Katrin Tomanek Algorithm Engineering Report TR07-2-013 December 2007 ISSN 1864-4503 Faculty of Computer Science Algorithm Engineering (Ls11) 44221 Dortmund / Germany http://ls11-www.cs.uni-dortmund.de/ Classical Probabilistic Models and Conditional Random Fields Roman Klinger∗ Katrin Tomanek∗ Fraunhofer Institute for Jena University Algorithms and Scientific Computing (SCAI) Language & Information Engineering (JULIE) Lab Schloss Birlinghoven F¨urstengraben 30 53754 Sankt Augustin, Germany 07743 Jena, Germany Dortmund University of Technology [email protected] Department of Computer Science Chair of Algorithm Engineering (Ls XI) 44221 Dortmund, Germany [email protected] [email protected] Contents 1 Introduction 2 2 Probabilistic Models 3 2.1 Na¨ıve Bayes ................................... 4 2.2 Hidden Markov Models ............................ 5 2.3 Maximum Entropy Model ........................... 6 3 Graphical Representation 10 3.1 Directed Graphical Models .......................... 12 3.2 Undirected Graphical Models ......................... 13 4 Conditional Random Fields 14 4.1 Basic Principles ................................. 14 4.2 Linear-chain CRFs ............................... 15 4.2.1 Training ................................. 19 4.2.2 Inference ................................ 23 4.3 Arbitrarily structured CRFs .......................... 24 4.3.1 Unrolling the graph .......................... 25 4.3.2 Training and Inference ......................... 26 5Summary 26 ∗ The authors contributed equally to this work. version 1.0 1 Introduction 1 Introduction Classification is known as the assignment of a class y to an observation x .This ∈Y ∈X task can be approached with probability theory by specifying a probability distribution to select the most likely class y for a given observation x. A well-known example (Russell and Norvig, 2003) is the classification of weather observations into categories, such as good or bad, = good, bad .
    [Show full text]