Dynamic Tree Block Coordinate Ascent
|
|
- Denis Todd
- 5 years ago
- Views:
Transcription
1 Dynamic Tree Block Coordinate Ascent Daniel Tarlow 1, Dhruv Batra 2 Pushmeet Kohli 3, Vladimir Kolmogorov 4 1: University of Toronto 3: Microso3 Research Cambridge 2: TTI Chicago 4: University College London InternaAonal Conference on Machine Learning (ICML), 2011
2 MAP in Large Discrete Models Many important problems can be expressed as a discrete Random Field (MRF, CRF) MAP inference is a fundamental problem min x X E(x) = min x X i V θ i (x i )+ (i,j) E θ ij (x i,x j ) InpainAng Stereo Object Class Labeling Protein Design / Side Chain PredicAon
3 Primal and Dual min x Primal A V E θ A (x A ) A V E Dual min x A θa (x A )= A V E h A Dual is a lower bound: less constrained version of primal is a reparameteriza)on, determined by messages h A* is height of unary or pairwise potenaal DefiniAon of reparameterizaaon: θ A (x A )= A V E A V E θ A (x A ) {x A } LP based message passing: find reparameterizaaon to maximize dual
4 Standard Linear Program based Message Passing Max Product Linear Programming (MPLP) Update edges in fixed order SequenAal Tree Reweighted Max Product (TRW S) SequenAally iterate over variables in fixed order Tree Block Coordinate Ascent (TBCA) [Sontag & Jaakkola, 2009] Update trees in fixed order Key: these are all energy oblivious Can we do be^er by being energy aware?
5 Example TBCA with StaJc Schedule: 630 messages needed TBCA with Dynamic Schedule: 276 messages needed
6 Benefit of Energy Awareness StaJc semngs Not all graph regions are equally difficult RepeaAng computaaon on easy parts is wasteful Easy region Dynamic semngs (e.g., learning, search) Harder region Small region of graph changes. ComputaAon on unchanged part is wasteful Unchanged Image Previous OpAmum Change Mask Changed
7 References and Related Work [Elidan et al., 2006], [Su^on & McCallum, 2007] Residual Belief PropagaAon. Pass most different messages first. [Chandrasekaran et al., 2007] Works only on conanuous variables. Very different formulaaon. [Batra et al., 2011] Local Primal Dual Gap for Tightening LP relaxaaons. [Kolmogorov, 2006] Weak Tree Agreement in relaaon to TRW S. [Sontag et al., 2009] Tree Block Coordinate Descent.
8 VisualizaJon of reparameterized energy States for each variable: red (R), green (G), or blue (B) "Good" local settings: (can assume "good" has cost 0, otherwise cost 1) G G G, B B G G G, B B G G B, B B B x 1 x 2 x 3 x 4 R G B
9 VisualizaJon of reparameterized energy States for each variable: red (R), green (G), or blue (B) "Good" local settings: R G B G G G, B B G G G, B B G G B, B B B "Don't be R or B" "Don't be R or G" x 1 x 2 x 3 x 4 "Don't be R or B" "Don't be R or B" "Don't be R or B" "Don't be R" HypotheJcal messages that e.g. residual max product would send.
10 VisualizaJon of reparameterized energy States for each variable: red (R), green (G), or blue (B) "Good" local settings: G G G, B B G G G, B B G G B, B B B x 1 x 2 x 3 x 4 R G B But we don't need to send any messages. We are at the global opjmum. Our scores (see later slides) are 0, so we wouldn't send any messages here.
11 VisualizaJon of reparameterized energy States for each variable: red (R), green (G), or blue (B) "Good" local settings: B G G, B B B G G, B B B G B, B B B x 1 x 2 x 3 x 4 R G B Change unary potenjals (e.g., during learning or search)
12 VisualizaJon of reparameterized energy States for each variable: red (R), green (G), or blue (B) "Good" local settings: B G G, B B B G G, B B B G B, B B B x 1 x 2 x 3 x 4 R G B Locally, best assignment for some variables change.
13 VisualizaJon of reparameterized energy States for each variable: red (R), green (G), or blue (B) "Good" local settings: R G B B G G, B B B G G, B B B G B, B B B "Don't be R or G" "Don't be R or G" x 1 x 2 x 3 x 4 "Don't be R or G" "Don't be R or G" "Don't be R or G" "Don't be R" HypotheJcal messages that e.g. residual max product would send.
14 VisualizaJon of reparameterized energy States for each variable: red (R), green (G), or blue (B) "Good" local settings: B G G, B B B G G, B B B G B, B B B x 1 x 2 x 3 x 4 R G B But we don't need to send any messages. We are at the global opjmum. Our scores (see later slides) are 0, so we wouldn't send any messages here.
15 VisualizaJon of reparameterized energy "Good" local settings: B G G, B B B G G, B B B G B, B B B x 1 x 2 x 3 x 4 Possible fix: look at how much sending messages on edge would improve dual. Would work in above case, but incorrectly ignores e.g. the subgraph below: "Good" local settings: B B B G, B G G R, G R R R R G B x 1 x 2 x 3 x 4
16 Key Slide B B B G, B B B R,G R R R x 1 x 2 x 3 x 4 Locally, everything looks opamal
17 Key Slide B B B G, B B B R,G R R R x 1 x 2 x 3 x 4 Try assigning a value to each variable
18 Key Slide Our main contribuaon Use primal (and dual) informajon to choose regions on which to pass messages B B B G, B B B R,G R R R x 1 x 2 x 3 x 4 SubopAmal SubopAmal Try assigning a value to each variable
19 Our FormulaJon Measure primal dual local agreement at edges and variables Local Primal Dual Gap (LPDG). Weak Tree Agreement (WTA). Choose forest with maximum disagreement Kruskal's algorithm, possibly terminated early Apply TBCA update on maximal trees Important! Minimize overhead. Use quanaaes that are already computed during inference, and carefully cache computaaons
20 Local Primal Dual Gap (LPDG) Score Difference between primal and dual objecaves Primal dual gap Given primal assignment x p and dual variables (messages) defining, primal dual gap is A V E θ A (x p A ) A V E min x A θa (x A ) = A V E primal dual ( ) θ A (x p A ) min θa (x A ) x A = A V E LPDG (A) Primal cost of node/edge Dual bound at node/edge e: local disagreement measure: e A = LPDG (A)
21 Shortcoming of LPDG Score: Loose RelaxaJons LPDG > 0, but dual opamal Filled circle means, black edge means
22 Weak Tree Agreement (WTA) [Kolmogorov 2006] Reparameterized potenaals are said to saasfy WTA if there exist non empty subsets for each node i D i X i such that θ i (x i )=h i min θij (x i,x j )=h ij x j D j x i D i x i D i, (i, j) E Filled circle means Black edge means labels labels At Weak Tree Agreement Not at Weak Tree Agreement
23 Weak Tree Agreement (WTA) [Kolmogorov 2006] Reparameterized potenaals are said to saasfy WTA if there exist non empty subsets for each node i D i X i such that θ i (x i )=h i min θij (x i,x j )=h ij x j D j x i D i x i D i, (i, j) E Filled circle means Black edge means labels labels At Weak Tree Agreement D 1 ={0} D 2 ={0,2} D 2 ={0,2} D 3 ={0} Not at Weak Tree Agreement D 1 ={0} D 2 ={2} D 2 ={0,2} D 3 ={0}
24 Weak Tree Agreement (WTA) [Kolmogorov 2006] Reparameterized potenaals are said to saasfy WTA if there exist non empty subsets for each node i D i X i such that θ i (x i )=h i min θij (x i,x j )=h ij x j D j x i D i x i D i, (i, j) E Filled circle means Black edge means labels labels At Weak Tree Agreement Not at Weak Tree Agreement D 1 ={0} D 2 ={2} D 2 ={0,2} D 3 ={0}
25 WTA Score e: local disagreement measure Costs: solid low do^ed medium else high D 2 ={0,2} D 3 ={0,2} e 23 = max( min(a,high), min(b,c) ) c Filled circle means, black edge means
26 WTA Score e: local disagreement measure Costs: solid low do^ed medium else high D 2 ={0,2} D 3 ={0,2} e 23 = max( min(a,high), min(b,c) ) c Filled circle means, black edge means
27 WTA Score e: local disagreement measure Costs: solid low do^ed medium else high D 2 ={0,2} D 3 ={0,2} e 23 = max a, c ( ) c = a c Filled circle means, black edge means
28 WTA Score e: local disagreement measure Costs: solid low do^ed medium else high D 2 ={0,2} D 3 ={0,2} e 23 = max a, c ( ) c = a c Filled circle means, black edge means
29 WTA Score e: local disagreement measure: node measure e i = max θ i (x i ) min θ i (x i ) x i D i x i
30 Single FormulaJon of LPDG and WTA Set a max history size parameter R. Store most recent R labelings of variable i in label set D i R=1: LPDG score. R>1: WTA score. Combine scores into undirected edge score:
31 ProperJes of LPDG/WTA Scores LPDG measure gives upper bound on possible dual improvement from passing messages on forest LPDG may overesamate "usefulness" of an edge e.g., on non Aght relaxaaons. LPDG > 0 WTA = 0 WTA measure addresses overesamate problem: is zero shortly a3er normal message passing would converge. Both only change when messages are passed on nearby region of graph.
32 Experiments Computer Vision: Stereo Image SegmentaAon Dynamic Image SegmentaAon Protein Design: StaAc problem CorrelaAon between measure and dual improvement Dynamic search applicaaon Algorithms TBCA: StaAc Schedule, LPDG Schedule, WTA Schedule MPLP [Sontag and Globerson implementaaon] TRW S [Kolmogorov ImplementaAon]
33 Experiments: Stereo 383x434 pixels, 16 labels. Po^s potenaals.
34 Experiments: Image SegmentaJon +,!" (, '(%& + (!&!* %*%, 5678,9:;03<.=0!&!)(!&!)!&!'(!&!', ->?@!AB5C ->?@!D-@ ->?@!?E/:2 -FD!G H@B!"" #"" $"" %"" -./0,12034.;2<(=>?541/@5 %*%$+ %*%$ %*%)+ %*%) ( 6ABC!DE.F 6ABC!G6C 6ABC!B9:>0 6HG!I JCE! " # $ %& %! -(./012345( :0 '(%& ) 375x500 pixels, 21 labels. General potenaals based on label co occurence.
35 Experiments: Dynamic Image SegmentaJon Sheep!&'# +,!"*!&'*!&') 6DEF!?6F,.A2B 6DEF!?6F,C7/:2B, Previous Opt Modify White Unaries ;357<=!&''!&'( Sheep!&'! New Opt Heatmap of Messages!&',!" #!" $!" % -,./012345, :0 Warm started DTBCA vs Warm started TRW S 375x500 pixels, 21 labels. Po^s potenaals.
36 Experiments: Dynamic Image SegmentaJon Airplane Previous Opt Modify White Unaries New Opt Heatmap of Messages Warm started DTBCA vs Warm started TRW S 375x500 pixels, 21 labels. Po^s potenaals.
37 Experiments: Protein Design Dual Improvement vs. Measure on Forest Dual Improvement 10 1 Random Blocks Max WTA Block Max LPDG Block 10 1 LPDG/WTA Measure Dual Improvement 10!3 10!5 Random Blocks Max LPDG Block 10!5 10!3 10!1 LPDG Measure Dual Improvement 10!3 10!5 Random Blocks Max WTA Block 10!5 10!3 10!1 WTA Measure Other protein experiments: (see paper) DTBCA vs. staac "stars" on small protein DTBCA converges to opamum in.39s vs TBCA in.86s SimulaAng node expansion in A* search on larger protein Similar dual for DTBCA in 5s as Warm started TRW S in 50s. Protein Design from Yanover et al.
38 Discussion Energy oblivious schedules can be wasteful. For LP based message passing, primal informaaon is useful for scheduling. We give two low overhead ways of including it Biggest win comes from dynamic applicaaons ExciAng future dynamic applicaaons: search, learning,...
39 Discussion Energy oblivious schedules can be wasteful. For LP based message passing, primal informaaon is useful for scheduling. We give two low overhead ways of including it Biggest win comes from dynamic applicaaons ExciAng future dynamic applicaaons: search, learning,... Thank You!
40 Unused slides
41 Schlesinger's Linear Program (LP) Real valued exact Marginal polytope approx LOCAL polytope (see next slide)
42 Primal Dual
43
44 WTA Score e: local disagreement measure D 2 ={0,2} D 3 ={0} Filled circle means, black edge means
An Introduction to LP Relaxations for MAP Inference
An Introduction to LP Relaxations for MAP Inference Adrian Weller MLSALT4 Lecture Feb 27, 2017 With thanks to David Sontag (NYU) for use of some of his slides and illustrations For more information, see
More informationTightening MRF Relaxations with Planar Subproblems
Tightening MRF Relaxations with Planar Subproblems Julian Yarkony, Ragib Morshed, Alexander T. Ihler, Charless C. Fowlkes Department of Computer Science University of California, Irvine {jyarkony,rmorshed,ihler,fowlkes}@ics.uci.edu
More informationConvergent message passing algorithms - a unifying view
Convergent message passing algorithms - a unifying view Talya Meltzer, Amir Globerson and Yair Weiss School of Computer Science and Engineering The Hebrew University of Jerusalem, Jerusalem, Israel {talyam,gamir,yweiss}@cs.huji.ac.il
More informationA discriminative view of MRF pre-processing algorithms supplementary material
A discriminative view of MRF pre-processing algorithms supplementary material Chen Wang 1,2 Charles Herrmann 2 Ramin Zabih 1,2 1 Google Research 2 Cornell University {chenwang, cih, rdz}@cs.cornell.edu
More informationModels for grids. Computer vision: models, learning and inference. Multi label Denoising. Binary Denoising. Denoising Goal.
Models for grids Computer vision: models, learning and inference Chapter 9 Graphical Models Consider models where one unknown world state at each pixel in the image takes the form of a grid. Loops in the
More informationPlanar Cycle Covering Graphs
Planar Cycle Covering Graphs Julian Yarkony, Alexander T. Ihler, Charless C. Fowlkes Department of Computer Science University of California, Irvine {jyarkony,ihler,fowlkes}@ics.uci.edu Abstract We describe
More informationThe More the Merrier: Parameter Learning for Graphical Models with Multiple MAPs
The More the Merrier: Parameter Learning for Graphical Models with Multiple MAPs Franziska Meier fmeier@usc.edu U. of Southern California, 94 Bloom Walk, Los Angeles, CA 989 USA Amir Globerson amir.globerson@mail.huji.ac.il
More informationReduce, Reuse & Recycle: Efficiently Solving Multi-Label MRFs
Reduce, Reuse & Recycle: Efficiently Solving Multi-Label MRFs Karteek Alahari 1 Pushmeet Kohli 2 Philip H. S. Torr 1 1 Oxford Brookes University 2 Microsoft Research, Cambridge Abstract In this paper,
More informationECE 6504: Advanced Topics in Machine Learning Probabilistic Graphical Models and Large-Scale Learning
ECE 6504: Advanced Topics in Machine Learning Probabilistic Graphical Models and Large-Scale Learning Topics Markov Random Fields: Inference Exact: VE Exact+Approximate: BP Readings: Barber 5 Dhruv Batra
More informationD-Separation. b) the arrows meet head-to-head at the node, and neither the node, nor any of its descendants, are in the set C.
D-Separation Say: A, B, and C are non-intersecting subsets of nodes in a directed graph. A path from A to B is blocked by C if it contains a node such that either a) the arrows on the path meet either
More informationCombinatorial Optimization
Combinatorial Optimization Frank de Zeeuw EPFL 2012 Today Introduction Graph problems - What combinatorial things will we be optimizing? Algorithms - What kind of solution are we looking for? Linear Programming
More informationMessage Passing Inference for Large Scale Graphical Models with High Order Potentials
Message Passing Inference for Large Scale Graphical Models with High Order Potentials Jian Zhang ETH Zurich jizhang@ethz.ch Alexander G. Schwing University of Toronto aschwing@cs.toronto.edu Raquel Urtasun
More informationRevisiting MAP Estimation, Message Passing and Perfect Graphs
James R. Foulds Nicholas Navaroli Padhraic Smyth Alexander Ihler Department of Computer Science University of California, Irvine {jfoulds,nnavarol,smyth,ihler}@ics.uci.edu Abstract Given a graphical model,
More informationImage Segmentation with a Bounding Box Prior Victor Lempitsky, Pushmeet Kohli, Carsten Rother, Toby Sharp Microsoft Research Cambridge
Image Segmentation with a Bounding Box Prior Victor Lempitsky, Pushmeet Kohli, Carsten Rother, Toby Sharp Microsoft Research Cambridge Dylan Rhodes and Jasper Lin 1 Presentation Overview Segmentation problem
More informationApproximate inference using planar graph decomposition
Approximate inference using planar graph decomposition Amir Globerson Tommi Jaakkola Computer Science and Artificial Intelligence Laboratory Massachusetts Institute of Technology Cambridge, MA 0239 gamir,tommi@csail.mit.edu
More informationInformation Processing Letters
Information Processing Letters 112 (2012) 449 456 Contents lists available at SciVerse ScienceDirect Information Processing Letters www.elsevier.com/locate/ipl Recursive sum product algorithm for generalized
More informationBelief propagation and MRF s
Belief propagation and MRF s Bill Freeman 6.869 March 7, 2011 1 1 Undirected graphical models A set of nodes joined by undirected edges. The graph makes conditional independencies explicit: If two nodes
More informationDynamic Planar-Cuts: Efficient Computation of Min-Marginals for Outer-Planar Models
Dynamic Planar-Cuts: Efficient Computation of Min-Marginals for Outer-Planar Models Dhruv Batra 1 1 Carnegie Mellon University www.ece.cmu.edu/ dbatra Tsuhan Chen 2,1 2 Cornell University tsuhan@ece.cornell.edu
More informationThe Lazy Flipper: Efficient Depth-limited Exhaustive Search in Discrete Graphical Models
The Lazy Flipper: Efficient Depth-limited Exhaustive Search in Discrete Graphical Models Bjoern Andres, Jörg H. Kappes, Thorsten Beier, Ullrich Köthe and Fred A. Hamprecht HCI, University of Heidelberg,
More informationApproximation Algorithms: The Primal-Dual Method. My T. Thai
Approximation Algorithms: The Primal-Dual Method My T. Thai 1 Overview of the Primal-Dual Method Consider the following primal program, called P: min st n c j x j j=1 n a ij x j b i j=1 x j 0 Then the
More informationOSU CS 536 Probabilistic Graphical Models. Loopy Belief Propagation and Clique Trees / Join Trees
OSU CS 536 Probabilistic Graphical Models Loopy Belief Propagation and Clique Trees / Join Trees Slides from Kevin Murphy s Graphical Model Tutorial (with minor changes) Reading: Koller and Friedman Ch
More informationMarkov Random Fields and Segmentation with Graph Cuts
Markov Random Fields and Segmentation with Graph Cuts Computer Vision Jia-Bin Huang, Virginia Tech Many slides from D. Hoiem Administrative stuffs Final project Proposal due Oct 27 (Thursday) HW 4 is out
More informationMRFs and Segmentation with Graph Cuts
02/24/10 MRFs and Segmentation with Graph Cuts Computer Vision CS 543 / ECE 549 University of Illinois Derek Hoiem Today s class Finish up EM MRFs w ij i Segmentation with Graph Cuts j EM Algorithm: Recap
More informationA Tutorial Introduction to Belief Propagation
A Tutorial Introduction to Belief Propagation James Coughlan August 2009 Table of Contents Introduction p. 3 MRFs, graphical models, factor graphs 5 BP 11 messages 16 belief 22 sum-product vs. max-product
More informationCAP5415-Computer Vision Lecture 13-Image/Video Segmentation Part II. Dr. Ulas Bagci
CAP-Computer Vision Lecture -Image/Video Segmentation Part II Dr. Ulas Bagci bagci@ucf.edu Labeling & Segmentation Labeling is a common way for modeling various computer vision problems (e.g. optical flow,
More informationEnergy Minimization Under Constraints on Label Counts
Energy Minimization Under Constraints on Label Counts Yongsub Lim 1, Kyomin Jung 1, and Pushmeet Kohli 2 1 Korea Advanced Istitute of Science and Technology, Daejeon, Korea yongsub@kaist.ac.kr, kyomin@kaist.edu
More informationSegmentation. Bottom up Segmentation Semantic Segmentation
Segmentation Bottom up Segmentation Semantic Segmentation Semantic Labeling of Street Scenes Ground Truth Labels 11 classes, almost all occur simultaneously, large changes in viewpoint, scale sky, road,
More informationMarkov/Conditional Random Fields, Graph Cut, and applications in Computer Vision
Markov/Conditional Random Fields, Graph Cut, and applications in Computer Vision Fuxin Li Slides and materials from Le Song, Tucker Hermans, Pawan Kumar, Carsten Rother, Peter Orchard, and others Recap:
More informationMarkov Networks in Computer Vision. Sargur Srihari
Markov Networks in Computer Vision Sargur srihari@cedar.buffalo.edu 1 Markov Networks for Computer Vision Important application area for MNs 1. Image segmentation 2. Removal of blur/noise 3. Stereo reconstruction
More informationLoopy Belief Propagation
Loopy Belief Propagation Research Exam Kristin Branson September 29, 2003 Loopy Belief Propagation p.1/73 Problem Formalization Reasoning about any real-world problem requires assumptions about the structure
More informationSupplementary Material: Decision Tree Fields
Supplementary Material: Decision Tree Fields Note, the supplementary material is not needed to understand the main paper. Sebastian Nowozin Sebastian.Nowozin@microsoft.com Toby Sharp toby.sharp@microsoft.com
More informationRounding-based Moves for Metric Labeling
Rounding-based Moves for Metric Labeling M. Pawan Kumar Ecole Centrale Paris & INRIA Saclay pawan.kumar@ecp.fr Abstract Metric labeling is a special case of energy minimization for pairwise Markov random
More informationConvex Max-Product Algorithms for Continuous MRFs with Applications to Protein Folding
Convex Max-Product Algorithms for Continuous MRFs with Applications to Protein Folding Jian Peng Tamir Hazan David McAllester Raquel Urtasun TTI Chicago, Chicago Abstract This paper investigates convex
More informationPairwise Matching through Max-Weight Bipartite Belief Propagation
Pairwise Matching through Max-Weight Bipartite Belief Propagation Zhen Zhang Qinfeng Shi 2 Julian McAuley 3 Wei Wei Yanning Zhang Anton van den Hengel 2 School of Computer Science and Engineering, Northwestern
More informationComputer Vision Group Prof. Daniel Cremers. 4. Probabilistic Graphical Models Directed Models
Prof. Daniel Cremers 4. Probabilistic Graphical Models Directed Models The Bayes Filter (Rep.) (Bayes) (Markov) (Tot. prob.) (Markov) (Markov) 2 Graphical Representation (Rep.) We can describe the overall
More informationBeyond Trees: MRF Inference via Outer-Planar Decomposition
Beyond Trees: MRF Inference via Outer-Planar Decomposition Dhruv Batra Carnegie Mellon Univerity www.ece.cmu.edu/ dbatra A. C. Gallagher Eastman Kodak Company andrew.c.gallagher@gmail.com Devi Parikh TTIC
More informationLearning and Recognizing Visual Object Categories Without First Detecting Features
Learning and Recognizing Visual Object Categories Without First Detecting Features Daniel Huttenlocher 2007 Joint work with D. Crandall and P. Felzenszwalb Object Category Recognition Generic classes rather
More informationComputer Vision Group Prof. Daniel Cremers. 4a. Inference in Graphical Models
Group Prof. Daniel Cremers 4a. Inference in Graphical Models Inference on a Chain (Rep.) The first values of µ α and µ β are: The partition function can be computed at any node: Overall, we have O(NK 2
More informationAnalysis: TextonBoost and Semantic Texton Forests. Daniel Munoz Februrary 9, 2009
Analysis: TextonBoost and Semantic Texton Forests Daniel Munoz 16-721 Februrary 9, 2009 Papers [shotton-eccv-06] J. Shotton, J. Winn, C. Rother, A. Criminisi, TextonBoost: Joint Appearance, Shape and Context
More informationOn Iteratively Constraining the Marginal Polytope for Approximate Inference and MAP
On Iteratively Constraining the Marginal Polytope for Approximate Inference and MAP David Sontag CSAIL Massachusetts Institute of Technology Cambridge, MA 02139 Tommi Jaakkola CSAIL Massachusetts Institute
More informationVariable Grouping for Energy Minimization
Variable Grouping for Energy Minimization Taesup Kim, Sebastian Nowozin, Pushmeet Kohli, Chang D. Yoo Dept. of Electrical Engineering, KAIST, Daejeon Korea Microsoft Research Cambridge, Cambridge UK kim.ts@kaist.ac.kr,
More informationComputer Vision Group Prof. Daniel Cremers. 4. Probabilistic Graphical Models Directed Models
Prof. Daniel Cremers 4. Probabilistic Graphical Models Directed Models The Bayes Filter (Rep.) (Bayes) (Markov) (Tot. prob.) (Markov) (Markov) 2 Graphical Representation (Rep.) We can describe the overall
More informationUndirected Graphical Models. Raul Queiroz Feitosa
Undirected Graphical Models Raul Queiroz Feitosa Pros and Cons Advantages of UGMs over DGMs UGMs are more natural for some domains (e.g. context-dependent entities) Discriminative UGMs (CRF) are better
More informationEfficiently Learning Random Fields for Stereo Vision with Sparse Message Passing
Efficiently Learning Random Fields for Stereo Vision with Sparse Message Passing Jerod J. Weinman 1, Lam Tran 2, and Christopher J. Pal 2 1 Dept. of Computer Science 2 Dept. of Computer Science Grinnell
More informationLinear programming and duality theory
Linear programming and duality theory Complements of Operations Research Giovanni Righini Linear Programming (LP) A linear program is defined by linear constraints, a linear objective function. Its variables
More informationMarkov Networks in Computer Vision
Markov Networks in Computer Vision Sargur Srihari srihari@cedar.buffalo.edu 1 Markov Networks for Computer Vision Some applications: 1. Image segmentation 2. Removal of blur/noise 3. Stereo reconstruction
More informationPart I: Sum Product Algorithm and (Loopy) Belief Propagation. What s wrong with VarElim. Forwards algorithm (filtering) Forwards-backwards algorithm
OU 56 Probabilistic Graphical Models Loopy Belief Propagation and lique Trees / Join Trees lides from Kevin Murphy s Graphical Model Tutorial (with minor changes) eading: Koller and Friedman h 0 Part I:
More informationESSP: An Efficient Approach to Minimizing Dense and Nonsubmodular Energy Functions
1 ESSP: An Efficient Approach to Minimizing Dense and Nonsubmodular Functions Wei Feng, Member, IEEE, Jiaya Jia, Senior Member, IEEE, and Zhi-Qiang Liu, arxiv:145.4583v1 [cs.cv] 19 May 214 Abstract Many
More informationLearning CRFs using Graph Cuts
Appears in European Conference on Computer Vision (ECCV) 2008 Learning CRFs using Graph Cuts Martin Szummer 1, Pushmeet Kohli 1, and Derek Hoiem 2 1 Microsoft Research, Cambridge CB3 0FB, United Kingdom
More informationNew Outer Bounds on the Marginal Polytope
New Outer Bounds on the Marginal Polytope David Sontag Tommi Jaakkola Computer Science and Artificial Intelligence Laboratory Massachusetts Institute of Technology Cambridge, MA 2139 dsontag,tommi@csail.mit.edu
More informationHeuristic Optimization Today: Linear Programming. Tobias Friedrich Chair for Algorithm Engineering Hasso Plattner Institute, Potsdam
Heuristic Optimization Today: Linear Programming Chair for Algorithm Engineering Hasso Plattner Institute, Potsdam Linear programming Let s first define it formally: A linear program is an optimization
More informationDynamic Planar-Cuts: Efficient Computation of Min-Marginals for Outer-Planar MRFs
Dynamic Planar-Cuts: Efficient Computation of Min-Marginals for Outer-Planar MRFs Dhruv Batra Carnegie Mellon University Tsuhan Chen Cornell University batradhruv@cmu.edu tsuhan@ece.cornell.edu Abstract
More informationParallel and Distributed Graph Cuts by Dual Decomposition
Parallel and Distributed Graph Cuts by Dual Decomposition Presented by Varun Gulshan 06 July, 2010 What is Dual Decomposition? What is Dual Decomposition? It is a technique for optimization: usually used
More informationMAP Estimation, Message Passing and Perfect Graphs
MAP Estimation, Message Passing and Perfect Graphs Tony Jebara November 25, 2009 1 Background Perfect Graphs Graphical Models 2 Matchings Bipartite Matching Generalized Matching 3 Perfect Graphs nand Markov
More informationPart II. C. M. Bishop PATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 8: GRAPHICAL MODELS
Part II C. M. Bishop PATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 8: GRAPHICAL MODELS Converting Directed to Undirected Graphs (1) Converting Directed to Undirected Graphs (2) Add extra links between
More informationMotivation: Shortcomings of Hidden Markov Model. Ko, Youngjoong. Solution: Maximum Entropy Markov Model (MEMM)
Motivation: Shortcomings of Hidden Markov Model Maximum Entropy Markov Models and Conditional Random Fields Ko, Youngjoong Dept. of Computer Engineering, Dong-A University Intelligent System Laboratory,
More informationEfficient Feature Learning Using Perturb-and-MAP
Efficient Feature Learning Using Perturb-and-MAP Ke Li, Kevin Swersky, Richard Zemel Dept. of Computer Science, University of Toronto {keli,kswersky,zemel}@cs.toronto.edu Abstract Perturb-and-MAP [1] is
More informationA MRF Shape Prior for Facade Parsing with Occlusions Supplementary Material
A MRF Shape Prior for Facade Parsing with Occlusions Supplementary Material Mateusz Koziński, Raghudeep Gadde, Sergey Zagoruyko, Guillaume Obozinski and Renaud Marlet Université Paris-Est, LIGM (UMR CNRS
More informationMeasuring Uncertainty in Graph Cut Solutions - Efficiently Computing Min-marginal Energies using Dynamic Graph Cuts
Measuring Uncertainty in Graph Cut Solutions - Efficiently Computing Min-marginal Energies using Dynamic Graph Cuts Pushmeet Kohli Philip H.S. Torr Department of Computing, Oxford Brookes University, Oxford.
More informationLecture 13 Segmentation and Scene Understanding Chris Choy, Ph.D. candidate Stanford Vision and Learning Lab (SVL)
Lecture 13 Segmentation and Scene Understanding Chris Choy, Ph.D. candidate Stanford Vision and Learning Lab (SVL) http://chrischoy.org Stanford CS231A 1 Understanding a Scene Objects Chairs, Cups, Tables,
More informationUncertainty Driven Multi-Scale Optimization
Uncertainty Driven Multi-Scale Optimization Pushmeet Kohli 1 Victor Lempitsky 2 Carsten Rother 1 1 Microsoft Research Cambridge 2 University of Oxford Abstract. This paper proposes a new multi-scale energy
More informationQuadratic Pseudo-Boolean Optimization(QPBO): Theory and Applications At-a-Glance
Quadratic Pseudo-Boolean Optimization(QPBO): Theory and Applications At-a-Glance Presented By: Ahmad Al-Kabbany Under the Supervision of: Prof.Eric Dubois 12 June 2012 Outline Introduction The limitations
More informationLocal cues and global constraints in image understanding
Local cues and global constraints in image understanding Olga Barinova Lomonosov Moscow State University *Many slides adopted from the courses of Anton Konushin Image understanding «To see means to know
More informationComparison of energy minimization algorithms for highly connected graphs
Comparison of energy minimization algorithms for highly connected graphs Vladimir Kolmogorov 1 and Carsten Rother 2 1 University College London; vnk@adastral.ucl.ac.uk 2 Microsoft Research Ltd., Cambridge,
More informationParallel constraint optimization algorithms for higher-order discrete graphical models: applications in hyperspectral imaging
Parallel constraint optimization algorithms for higher-order discrete graphical models: applications in hyperspectral imaging MSc Billy Braithwaite Supervisors: Prof. Pekka Neittaanmäki and Phd Ilkka Pölönen
More informationEfficient AND/OR Search Algorithms for Exact MAP Inference Task over Graphical Models
Efficient AND/OR Search Algorithms for Exact MAP Inference Task over Graphical Models Akihiro Kishimoto IBM Research, Ireland Joint work with Radu Marinescu and Adi Botea Outline 1 Background 2 RBFAOO
More informationMarkov Random Fields and Segmentation with Graph Cuts
Markov Random Fields and Segmentation with Graph Cuts Computer Vision Jia-Bin Huang, Virginia Tech Many slides from D. Hoiem Administrative stuffs Final project Proposal due Oct 30 (Monday) HW 4 is out
More informationChapter 8 of Bishop's Book: Graphical Models
Chapter 8 of Bishop's Book: Graphical Models Review of Probability Probability density over possible values of x Used to find probability of x falling in some range For continuous variables, the probability
More informationBeyond Pairwise Energies: Efficient Optimization for Higher-order MRFs
Beyond Pairwise Energies: Efficient Optimization for Higher-order MRFs Nikos Komodakis Computer Science Department, University of Crete komod@csd.uoc.gr Nikos Paragios Ecole Centrale de Paris/INRIA Saclay
More informationCRFs for Image Classification
CRFs for Image Classification Devi Parikh and Dhruv Batra Carnegie Mellon University Pittsburgh, PA 15213 {dparikh,dbatra}@ece.cmu.edu Abstract We use Conditional Random Fields (CRFs) to classify regions
More informationCurvature Prior for MRF-based Segmentation and Shape Inpainting
Curvature Prior for MRF-based Segmentation and Shape Inpainting Alexander Shekhovtsov, Pushmeet Kohli +, Carsten Rother + Center for Machine Perception, Czech Technical University, Microsoft Research Cambridge,
More informationGetting Feasible Variable Estimates From Infeasible Ones: MRF Local Polytope Study
2013 IEEE International Conference on Computer Vision Workshops Getting Feasible Variable Estimates From Infeasible Ones: MRF Local Polytope Study Bogdan Savchynskyy University of Heidelberg, Germany bogdan.savchynskyy@iwr.uni-heidelberg.de
More informationA Comparative Study of Modern Inference Techniques for Structured Discrete Energy Minimization Problems
A Comparative Study of Modern Inference Techniques for Structured Discrete Energy Minimization Problems Jörg H. Kappes Bjoern Andres Fred A. Hamprecht Christoph Schnörr Sebastian Nowozin Dhruv Batra Sungwoong
More informationHigher-order models in Computer Vision
Chapter 1 Higher-order models in Computer Vision PUSHMEET KOHLI Machine Learning and Perception Microsoft Research Cambridge, UK Email: pkohli@microsoft.com CARSTEN ROTHER Machine Learning and Perception
More informationDiverse M-Best Solutions in Markov Random Fields
Diverse M-Best Solutions in Markov Random Fields Dhruv Batra 1, Payman Yadollahpour 1, Abner Guzman-Rivera 2, and Gregory Shakhnarovich 1 1 TTI-Chicago 2 UIUC Abstract. Much effort has been directed at
More informationSTA 4273H: Statistical Machine Learning
STA 4273H: Statistical Machine Learning Russ Salakhutdinov Department of Statistics! rsalakhu@utstat.toronto.edu! http://www.utstat.utoronto.ca/~rsalakhu/ Sidney Smith Hall, Room 6002 Lecture 5 Inference
More informationIntroduction to Optimization
Introduction to Optimization Second Order Optimization Methods Marc Toussaint U Stuttgart Planned Outline Gradient-based optimization (1st order methods) plain grad., steepest descent, conjugate grad.,
More informationWinning the PASCAL 2011 MAP Challenge with Enhanced AND/OR Branch-and-Bound
Winning the PASCAL 2011 MAP Challenge with Enhanced AND/OR Branch-and-Bound Lars Otten, Alexander Ihler, Kalev Kask, Rina Dechter Department of Computer Science University of California, Irvine, U.S.A.
More informationPATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 8: GRAPHICAL MODELS
PATTERN RECOGNITION AND MACHINE LEARNING CHAPTER 8: GRAPHICAL MODELS Bayesian Networks Directed Acyclic Graph (DAG) Bayesian Networks General Factorization Bayesian Curve Fitting (1) Polynomial Bayesian
More informationParallel and Distributed Vision Algorithms Using Dual Decomposition
Parallel and Distributed Vision Algorithms Using Dual Decomposition Petter Strandmark, Fredrik Kahl, Thomas Schoenemann Centre for Mathematical Sciences Lund University, Sweden Abstract We investigate
More informationNetwork Design and Optimization course
Effective maximum flow algorithms Modeling with flows Network Design and Optimization course Lecture 5 Alberto Ceselli alberto.ceselli@unimi.it Dipartimento di Tecnologie dell Informazione Università degli
More information1. Lecture notes on bipartite matching
Massachusetts Institute of Technology 18.453: Combinatorial Optimization Michel X. Goemans February 5, 2017 1. Lecture notes on bipartite matching Matching problems are among the fundamental problems in
More informationPackage CRF. February 1, 2017
Version 0.3-14 Title Conditional Random Fields Package CRF February 1, 2017 Implements modeling and computational tools for conditional random fields (CRF) model as well as other probabilistic undirected
More informationClassifier Case Study: Viola-Jones Face Detector
Classifier Case Study: Viola-Jones Face Detector P. Viola and M. Jones. Rapid object detection using a boosted cascade of simple features. CVPR 2001. P. Viola and M. Jones. Robust real-time face detection.
More informationProbabilistic Graphical Models
School of Computer Science Probabilistic Graphical Models Theory of Variational Inference: Inner and Outer Approximation Eric Xing Lecture 14, February 29, 2016 Reading: W & J Book Chapters Eric Xing @
More informationCS246: Mining Massive Datasets Jure Leskovec, Stanford University
CS246: Mining Massive Datasets Jure Leskovec, Stanford University http://cs246.stanford.edu [Kumar et al. 99] 2/13/2013 Jure Leskovec, Stanford CS246: Mining Massive Datasets, http://cs246.stanford.edu
More informationJoint optimization of segmentation and appearance models November Vicente, 26, Kolmogorov, / Rothe 14
Joint optimization of segmentation and appearance models Vicente, Kolmogorov, Rother Chris Russell November 26, 2009 Joint optimization of segmentation and appearance models November Vicente, 26, Kolmogorov,
More informationSet 5: Constraint Satisfaction Problems
Set 5: Constraint Satisfaction Problems ICS 271 Fall 2014 Kalev Kask ICS-271:Notes 5: 1 The constraint network model Outline Variables, domains, constraints, constraint graph, solutions Examples: graph-coloring,
More informationA Dendrogram. Bioinformatics (Lec 17)
A Dendrogram 3/15/05 1 Hierarchical Clustering [Johnson, SC, 1967] Given n points in R d, compute the distance between every pair of points While (not done) Pick closest pair of points s i and s j and
More informationVisual Recognition: Examples of Graphical Models
Visual Recognition: Examples of Graphical Models Raquel Urtasun TTI Chicago March 6, 2012 Raquel Urtasun (TTI-C) Visual Recognition March 6, 2012 1 / 64 Graphical models Applications Representation Inference
More informationHow Hard Is Inference for Structured Prediction?
How Hard Is Inference for Structured Prediction? Tim Roughgarden (Stanford University) joint work with Amir Globerson (Tel Aviv), David Sontag (NYU), and Cafer Yildirum (NYU) 1 Structured Prediction structured
More informationIn this lecture, we ll look at applications of duality to three problems:
Lecture 7 Duality Applications (Part II) In this lecture, we ll look at applications of duality to three problems: 1. Finding maximum spanning trees (MST). We know that Kruskal s algorithm finds this,
More informationBig and Tall: Large Margin Learning with High Order Losses
Big and Tall: Large Margin Learning with High Order Losses Daniel Tarlow University of Toronto dtarlow@cs.toronto.edu Richard Zemel University of Toronto zemel@cs.toronto.edu Abstract Graphical models
More informationarxiv: v3 [cs.ai] 22 Dec 2015
Diffusion Methods for Classification with Pairwise Relationships arxiv:55.672v3 [cs.ai] 22 Dec 25 Pedro F. Felzenszwalb Brown University Providence, RI, USA pff@brown.edu December 23, 25 Abstract Benar
More informationDESIGN AND ANALYSIS OF ALGORITHMS GREEDY METHOD
1 DESIGN AND ANALYSIS OF ALGORITHMS UNIT II Objectives GREEDY METHOD Explain and detail about greedy method Explain the concept of knapsack problem and solve the problems in knapsack Discuss the applications
More informationAn LP View of the M-best MAP problem
An LP View of the M-best MAP problem Menachem Fromer Amir Globerson School of Computer Science and Engineering The Hebrew University of Jerusalem {fromer,gamir}@cs.huji.ac.il Abstract We consider the problem
More informationLocal Search. (Textbook Chpt 4.8) Computer Science cpsc322, Lecture 14. May, 30, CPSC 322, Lecture 14 Slide 1
Local Search Computer Science cpsc322, Lecture 14 (Textbook Chpt 4.8) May, 30, 2017 CPSC 322, Lecture 14 Slide 1 Announcements Assignment1 due now! Assignment2 out today CPSC 322, Lecture 10 Slide 2 Lecture
More informationStructured Learning. Jun Zhu
Structured Learning Jun Zhu Supervised learning Given a set of I.I.D. training samples Learn a prediction function b r a c e Supervised learning (cont d) Many different choices Logistic Regression Maximum
More informationSemi-Global Matching: a principled derivation in terms of Message Passing
Semi-Global Matching: a principled derivation in terms of Message Passing Amnon Drory 1, Carsten Haubold 2, Shai Avidan 1, Fred A. Hamprecht 2 1 Tel Aviv University, 2 University of Heidelberg Semi-global
More informationGeometric Registration for Deformable Shapes 3.3 Advanced Global Matching
Geometric Registration for Deformable Shapes 3.3 Advanced Global Matching Correlated Correspondences [ASP*04] A Complete Registration System [HAW*08] In this session Advanced Global Matching Some practical
More information