1 MPE and Partial Inversion in Lifted Probabilistic Variable Elimination Rodrigo de Salvo Braz University of Illinois at Urbana-Champaign with Eyal Amir.

Slides:



Advertisements
Similar presentations
TWO STEP EQUATIONS 1. SOLVE FOR X 2. DO THE ADDITION STEP FIRST
Advertisements

Advanced Piloting Cruise Plot.
Feichter_DPG-SYKL03_Bild-01. Feichter_DPG-SYKL03_Bild-02.
Copyright © 2003 Pearson Education, Inc. Slide 1 Computer Systems Organization & Architecture Chapters 8-12 John D. Carpinelli.
Chapter 1 The Study of Body Function Image PowerPoint
Copyright © 2011, Elsevier Inc. All rights reserved. Chapter 6 Author: Julia Richards and R. Scott Hawley.
Author: Julia Richards and R. Scott Hawley
1 Copyright © 2013 Elsevier Inc. All rights reserved. Appendix 01.
Properties Use, share, or modify this drill on mathematic properties. There is too much material for a single class, so you’ll have to select for your.
UNITED NATIONS Shipment Details Report – January 2006.
1 RA I Sub-Regional Training Seminar on CLIMAT&CLIMAT TEMP Reporting Casablanca, Morocco, 20 – 22 December 2005 Status of observing programmes in RA I.
Jeopardy Q 1 Q 6 Q 11 Q 16 Q 21 Q 2 Q 7 Q 12 Q 17 Q 22 Q 3 Q 8 Q 13
Jeopardy Q 1 Q 6 Q 11 Q 16 Q 21 Q 2 Q 7 Q 12 Q 17 Q 22 Q 3 Q 8 Q 13
Properties of Real Numbers CommutativeAssociativeDistributive Identity + × Inverse + ×
FACTORING ax2 + bx + c Think “unfoil” Work down, Show all steps.
Year 6 mental test 5 second questions
Year 6 mental test 10 second questions
Solve Multi-step Equations
REVIEW: Arthropod ID. 1. Name the subphylum. 2. Name the subphylum. 3. Name the order.
Break Time Remaining 10:00.
ABC Technology Project
1 Undirected Breadth First Search F A BCG DE H 2 F A BCG DE H Queue: A get Undiscovered Fringe Finished Active 0 distance from A visit(A)
2 |SharePoint Saturday New York City
Green Eggs and Ham.
VOORBLAD.
15. Oktober Oktober Oktober 2012.
1 Breadth First Search s s Undiscovered Discovered Finished Queue: s Top of queue 2 1 Shortest path from s.
1 RA III - Regional Training Seminar on CLIMAT&CLIMAT TEMP Reporting Buenos Aires, Argentina, 25 – 27 October 2006 Status of observing programmes in RA.
BIOLOGY AUGUST 2013 OPENING ASSIGNMENTS. AUGUST 7, 2013  Question goes here!
Factor P 16 8(8-5ab) 4(d² + 4) 3rs(2r – s) 15cd(1 + 2cd) 8(4a² + 3b²)
Basel-ICU-Journal Challenge18/20/ Basel-ICU-Journal Challenge8/20/2014.
1..
© 2012 National Heart Foundation of Australia. Slide 2.
Lets play bingo!!. Calculate: MEAN Calculate: MEDIAN
LO: Count up to 100 objects by grouping them and counting in 5s 10s and 2s. Mrs Criddle: Westfield Middle School.
Understanding Generalist Practice, 5e, Kirst-Ashman/Hull
Chapter 5 Test Review Sections 5-1 through 5-4.
Addition 1’s to 20.
Model and Relationships 6 M 1 M M M M M M M M M M M M M M M M
25 seconds left…...
Equal or Not. Equal or Not
Slippery Slope
H to shape fully developed personality to shape fully developed personality for successful application in life for successful.
Januar MDMDFSSMDMDFSSS
Week 1.
Statistical Inferences Based on Two Samples
Analyzing Genes and Genomes
We will resume in: 25 Minutes.
©Brooks/Cole, 2001 Chapter 12 Derived Types-- Enumerated, Structure and Union.
Essential Cell Biology
Clock will move after 1 minute
Intracellular Compartments and Transport
PSSA Preparation.
Immunobiology: The Immune System in Health & Disease Sixth Edition
Essential Cell Biology
Immunobiology: The Immune System in Health & Disease Sixth Edition
Immunobiology: The Immune System in Health & Disease Sixth Edition
Energy Generation in Mitochondria and Chlorplasts
Select a time to count down from the clock above
Murach’s OS/390 and z/OS JCLChapter 16, Slide 1 © 2002, Mike Murach & Associates, Inc.
1 MPE and Partial Inversion in Lifted Probabilistic Variable Elimination Rodrigo de Salvo Braz University of Illinois at Urbana-Champaign with Eyal Amir.
1 MPE and Partial Inversion in Lifted Probabilistic Variable Elimination Rodrigo de Salvo Braz University of Illinois at Urbana-Champaign with Eyal Amir.
Lifted First-Order Probabilistic Inference Rodrigo de Salvo Braz SRI International joint work with Eyal Amir and Dan Roth.
1 Lifted First-Order Probabilistic Inference Rodrigo de Salvo Braz University of Illinois at Urbana-Champaign with Eyal Amir and Dan Roth.
Presentation transcript:

1 MPE and Partial Inversion in Lifted Probabilistic Variable Elimination Rodrigo de Salvo Braz University of Illinois at Urbana-Champaign with Eyal Amir and Dan Roth

Page 2 Lifted Probabilistic Inference We assume probabilistic statements such as 8 Person, Disease P(sick(Person,Disease) | epidemics(Disease)) = 0.3 Typical approach is grounding. We seek to do inference at first-order level, like it is done in logic. Faster and more intelligible. Two contributions: Partial inversion: more general technique than previous work (IJCAI '05) MPE and Lifted assignments

Page 3 Representing structure sick(mary,measles) epidemic(measles)epidemic(flu) sick(mary,flu) … … sick(bob,measles)sick(bob,flu) …… …… sick(P,D) epidemic(D) Poole (2003) named these parfactors, for parameterized factors Atom Logical variable

Page 4 Parfactor sick(Person,Disease) epidemic(Disease) 8 Person, Disease sick(Person,Disease), epidemic(Disease))

Page 5 Parfactor sick(Person,Disease) epidemic(Disease) 8 Person, Disease sick(Person,Disease), epidemic(Disease)), Person mary, Disease flu

Page 6 Joint Distribution As in propositional case, proportional to product of all factors But here, all factors means all instantiations of all parfactors: P(...) X (p(X)) X,Y (p(X),q(X,Y))

Page 7 Inference task - Marginalization q(X,Y) X (p(X)) X,Y (p(X),q(X,Y)) Marginal on all random variables in p(X): summation over all assignments to all instances of q(X,Y)

Page 8 The FOVE Algorithm First-Order Variable Elimination (FOVE): a generalization of Variable Elimination in propositional graphical models. Eliminates classes of random variables at once.

Page 9 FOVE P(hospital(mary)) = ? sick(mary,measles) hospital(mary) sick(mary, D) D measles epidemic(measles)epidemic(D) D measles

Page 10 FOVE P(hospital(mary)) = ? sick(mary,measles) hospital(mary) sick(mary, D) D measles epidemic(D) D measles

Page 11 FOVE hospital(mary) sick(mary, D) D measles epidemic(D) D measles P(hospital(mary)) = ?

Page 12 FOVE P(hospital(mary)) = ? hospital(mary) sick(mary, D) D measles

Page 13 FOVE P(hospital(mary)) = ? hospital(mary)

Page 14 e(D) D1 D2 (e(D 1 ),e(D 2 )) = e(D) (0,0) #(0,0) in assignment (0,1) #(0,1) in assignment (1,0) #(1,0) in assignment (1,1) #(1,1) in assignment Let i be the number of e(D)s assigned 1: = i v1,v2 (v1,v2) #(v1,v2) given i (number of assignments with |{D : e(D)=1}| = i) Counting Elimination - A Combinatorial Approach

Page 15 It does not work on eliminating class epidemic from (epidemic(D 1, Region), epidemic(D 2, Region), donations). In general, counting elimination does not apply when atoms share logical variables. Here, Region is shared between atoms. Counting Elimination - Conditions

Page 16 Partial Inversion Provides a way of not sharing logical variables e(D,R) D1 D2,R e(D1,R), e(D2,R), d ) R e(D,r) D1 D2 e(D1,r), e(D2,r), d ) ( R is now bound, so not a variable anymore) R d ) = d ) |R| = d )

Page 17 Partial Inversion, graphically epidemic(D2,r 1 ) epidemic(D1,r 1 ) D1 D2 donations epidemic(D2,R) epidemic(D1,R) D1 D2 donations epidemic(D2,r 10 ) epidemic(D1,r 10 ) D1 D2 … … Each instance a counting elimination problem

Page 18 Another (not so partial) inversion q(X,Y) X,Y (p(X),q(X,Y)) (expensive) = X,Y q(X,Y) (p(X),q(X,Y)) (propositional) = X,Y '(p(X)) = X ' Y (p(X)) = X ''(p(X)) (marginal on p(X) )

Page 19 Another (not so partial) inversion … q(x 1,y 1 ) p(x 1 ) q(x n,y n ) p(x n ) … q(X,Y) p(X) Each instance a propositional elimination problem

Page 20 Partial inversion conditions friends(X,Y), friends(Y,X)) Cannot partially invert on X,Y because friends(bob,mary) appears in more than one instance of parfactor. friends(mary,bob) friends(bob,mary) friends(Y,X) friends(X,Y) friends(bob,mary) … X Y friends(mary,bob) …

Page 21 Summary of Partial Inversion More general than previous Inversion Elimination. Generates Counting Elimination or Propositional sub-problems. Cannot be applied to entangled parfactors. Does not depend on domain size.

Page 22 Second contribution: Lifted MPE In propositional case, MPE done by factors containing MPE of eliminated variables. AB C D

Page 23 MPE AB D BD MPE 000.3C= C= C= C=1 In propositional case, MPE done by factors containing MPE of eliminated variables.

Page 24 MPE AB B MPE 00.5C=1,D=0 11.4C=1,D=1 In propositional case, MPE done by factors containing MPE of eliminated variables.

Page 25 MPE A A MPE(B,C,D) 00.9B=0,C=1,D=0 10.7B=1,C=1,D=1 In propositional case, MPE done by factors containing MPE of eliminated variables.

Page 26 MPE MPE 0.9A=0,B=1,C=1,D=1 In propositional case, MPE done by factors containing MPE of eliminated variables.

Page 27 MPE Same idea in First-order case But factors are quantified and so are assignments: p(X)q(X,Y) MPE r(X,Y) = r(X,Y) = r(X,Y) = r(X,Y) = 1 8 X, Y (p(X), q(X,Y))

Page 28 MPE After Inversion Elimination of q(X,Y): p(X)q(X,Y) MPE r(X,Y) = r(X,Y) = r(X,Y) = r(X,Y) = 1 8 X, Y (p(X), q(X,Y)) p(X) MPE Y q(X,Y) = 1, r(X,Y) = Y q(X,Y) = 0, r(X,Y) = 1 8 X (p(X)) Lifted assignments

Page 29 MPE After Inversion Elimination of p(X): 8 X (p(X)) MPE X 8 Y p(X) = 0, q(X,Y) = 1, r(X,Y) = 0 () p(X) MPE Y q(X,Y) = 1, r(X,Y) = Y q(X,Y) = 0, r(X,Y) = 1

Page 30 MPE After Counting Elimination of e: e(D1)e(D2) MPE r(D1,D2) = r(D1,D2) = r(D1,D2) = r(D1,D2) = 1 8 D1, D2 (e(D1), e(D2)) MPE D1,D2 e(D1)=0, e(D2) = 0, r(D1,D2) = D1,D2 e(D1)=0, e(D2) = 1, r(D1,D2) = D1,D2 e(D1)=1, e(D2) = 0, r(D1,D2) = D1,D2 e(D1)=1, e(D2) = 1, r(D1,D2) = 1 ()

Page 31 Conclusions Partial Inversion: More general algorithm, subsumes Inversion elimination Lifted Most Probable Explanation (MPE) same idea as in propositional VE, but with Lifted assignments: describe sets of basic assignments universally quantified comes from Partial Inversion existentially quantified comes from Counting elimination Ultimate goal: to perform lifted probabilistic inference in way similar to logic inference: without grounding and at a higher level.

Page 32