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

ECE Synthesis & Verification - L271 ECE 697B (667) Spring 2006 Synthesis and Verification of Digital Systems Model Checking basics.
1 MPE and Partial Inversion in Lifted Probabilistic Variable Elimination Rodrigo de Salvo Braz University of Illinois at Urbana-Champaign with Eyal Amir.
Propositions and Connectives Conditionals and Bi-conditionals Quantifiers.
First Order Predicate Logic to English Translation.
Lifted First-Order Probabilistic Inference Rodrigo de Salvo Braz, Eyal Amir and Dan Roth This research is supported by ARDA’s AQUAINT Program, by NSF grant.
Plan Recognition with Multi- Entity Bayesian Networks Kathryn Blackmond Laskey Department of Systems Engineering and Operations Research George Mason University.
Natural Language Processing  First-Order Predicate Logic - Predicates  Short Introduction to Description Logics  PowerLoom.
Recursive Random Fields Daniel Lowd University of Washington June 29th, 2006 (Joint work with Pedro Domingos)
Discrete Mathematics Goals of a Discrete Mathematics Learn how to think mathematically 1. Mathematical Reasoning Foundation for discussions of methods.
CSCI 115 Chapter 2 Logic. CSCI 115 §2.1 Propositions and Logical Operations.
Relational Probability Models Brian Milch MIT 9.66 November 27, 2007.
Inference in First-Order logic Department of Computer Science & Engineering Indian Institute of Technology Kharagpur.
1 Inductive Logic Programming. Part 2 Based partially on Luc De Raedt’s slides
Pairwise Document Similarity in Large Collections with MapReduce Tamer Elsayed, Jimmy Lin, and Douglas W. Oard Association for Computational Linguistics,
Chapter 1 Logic Section 1-1 Statements Open your book to page 1 and read the section titled “To the Student” Now turn to page 3 where we will read the.
Chapter 1, Part II: Predicate Logic With Question/Answer Animations.
Christopher Re and Dan Suciu University of Washington Efficient Evaluation of HAVING Queries on a Probabilistic Database.
The Bernays-Schönfinkel Fragment of First-Order Autoepistemic Logic Peter Baumgartner MPI Informatik, Saarbrücken.
2.3Logical Implication: Rules of Inference From the notion of a valid argument, we begin a formal study of what we shall mean by an argument and when such.
ARTIFICIAL INTELLIGENCE [INTELLIGENT AGENTS PARADIGM] Professor Janis Grundspenkis Riga Technical University Faculty of Computer Science and Information.
© Jalal Kawash 2010Peeking into Computer Science George Boole English Mathematician His The Mathematical Analysis of Logic, 1848 is the first.
Chapter 1, Part II: Predicate Logic With Question/Answer Animations.
Unification Algorithm Input: a finite set Σ of simple expressions Output: a mgu for Σ (if Σ is unifiable) 1. Set k = 0 and  0 = . 2. If Σ  k is a singleton,
Lifted First-Order Probabilistic Inference Rodrigo de Salvo Braz SRI International joint work with Eyal Amir and Dan Roth.
Markov Logic Networks Pedro Domingos Dept. Computer Science & Eng. University of Washington (Joint work with Matt Richardson)
1 Predicate (Relational) Logic 1. Introduction The propositional logic is not powerful enough to express certain types of relationship between propositions.
07 KM Page 1 ECEn/CS 224 Karnaugh Maps. 07 KM Page 2 ECEn/CS 224 What are Karnaugh Maps? A simpler way to handle most (but not all) jobs of manipulating.
Fundamentals of Logic 1. What is a valid argument or proof? 2. Study system of logic 3. In proving theorems or solving problems, creativity and insight.
Ahmad Almulhem, KFUPM 2010 COE 202: Digital Logic Design Combinational Logic Part 3 Dr. Ahmad Almulhem ahmadsm AT kfupm Phone: Office:
1 Lifted First-Order Probabilistic Inference Rodrigo de Salvo Braz University of Illinois at Urbana-Champaign with Eyal Amir and Dan Roth.
(CSC 102) Lecture 8 Discrete Structures. Previous Lectures Summary Predicates Set Notation Universal and Existential Statement Translating between formal.
Dr. Muhammed Al-Mulhem ICS An Introduction to Logical Programming.
1 Sections 1.3 and 1.4 Predicates & Quantifiers. 2 Propositional Functions In a mathematical assertion, such as x < 3, there are two parts: –the subject,
1 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Lecture 02: QUANTIFIERS Sections 1.3 and 1.4 Jarek Rossignac CS1050:
Fall 2008/2009 I. Arwa Linjawi & I. Asma’a Ashenkity 1 The Foundations: Logic and Proofs Predicates and Quantifiers.
Discrete Structures – CNS 2300
CS 285- Discrete Mathematics Lecture 4. Section 1.3 Predicate logic Predicate logic is an extension of propositional logic that permits concisely reasoning.
For Friday Read chapter 8 Homework: –Chapter 7, exercises 2 and 10 Program 1, Milestone 2 due.
Instructor: Eyal Amir Grad TAs: Wen Pu, Yonatan Bisk Undergrad TAs: Sam Johnson, Nikhil Johri CS 440 / ECE 448 Introduction to Artificial Intelligence.
Chapter 2 Fundamentals of Logic 1. What is a valid argument or proof?
CHAPTER 1 INTRODUCTION TO DIGITAL LOGIC
First-Order Probabilistic Inference Rodrigo de Salvo Braz.
1-1 An Introduction to Logical Programming Sept
Discrete Mathematics Mathematical reasoning: think logically; know how to prove Combinatorial analysis: know how to count Discrete structures: represent.
Instructor: Eyal Amir Grad TAs: Wen Pu, Yonatan Bisk Undergrad TAs: Sam Johnson, Nikhil Johri CS 440 / ECE 448 Introduction to Artificial Intelligence.
PREDICATES AND QUANTIFIERS COSC-1321 Discrete Structures 1.
CHAPTER 1 INTRODUCTION TO DIGITAL LOGIC. De Morgan’s Theorem De Morgan’s Theorem.
DE MORGAN’S THEOREM. De Morgan’s Theorem De Morgan’s Theorem.
Page 1 July 2008 ICML Workshop on Prior Knowledge for Text and Language Constraints as Prior Knowledge Ming-Wei Chang, Lev Ratinov, Dan Roth Department.
First-Order Probabilistic Inference Rodrigo de Salvo Braz SRI International.
1 Variable Elimination Graphical Models – Carlos Guestrin Carnegie Mellon University October 15 th, 2008 Readings: K&F: 8.1, 8.2, 8.3,
A Structural Characterization of Temporal Dynamic Controllability Paul Morris NASA Ames Research Center Moffett Field, CA 94035, U.S.A.
Daphne Koller Variable Elimination Variable Elimination Algorithm Probabilistic Graphical Models Inference.
CS498-EA Reasoning in AI Lecture #19 Professor: Eyal Amir Fall Semester 2011.
De Morgan’s Theorem,.
Intelligent Systems (AI-2) Computer Science cpsc422, Lecture 30
اختر أي شخصية واجعلها تطير!
Lifted First-Order Probabilistic Inference [de Salvo Braz, Amir, and Roth, 2005] Daniel Lowd 5/11/2005.
Anytime Lifted Belief Propagation
Combinatorial Logic Circuit
COE 202: Digital Logic Design Combinational Logic Part 3
Discrete Mathematics Lecture 4 & 5: Predicate and Quantifier
Lecture 07: E/R Diagrams and Functional Dependencies
Discrete Mathematics Lecture 3: Predicate and Quantifier
Automatic Inference in PyBLOG
Discrete Mathematics Lecture 4 & 5: Predicate and Quantifier
Clique Tree Algorithm: Computation
First Order Probabilistic Inference, by David Poole (2003)
Ensuring Security in Critical Infrastructures with Probabilistic Relational Modeling Part 2 Summer School “Hot Topics in Cyber Security” Rome Ralf Möller.
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 This talk Lifted Probabilistic Inference: performing probabilistic inference from a first-order level Two contributions: Partial inversion: more general algorithm compared to 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 Lifted Probabilistic Inference First-Order Variable Elimination (FOVE): a generalization of Variable Elimination in propositional graphical models. Eliminates classes of random variables at once.

Page 7 FOVE P(hospital(mary) | sick(mary, measles)) = ? sick(mary,measles) hospital(mary) sick(mary, D) D  measles epidemic(measles)epidemic(D) D  measles

Page 8 FOVE P(hospital(mary) | sick(mary, measles)) = ? sick(mary,measles) hospital(mary) sick(mary, D) D  measles epidemic(D) D  measles

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

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

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

Page 12 FOVE Previously, two operations: Inversion elimination Counting elimination

Page 13 Inversion Elimination  q(X,Y)  X,Y  (p(X),q(X,Y)) =  X,Y  q(X,Y)  (p(X),q(X,Y)) =  X,Y  '(p(X)) =  X  '  Y  (p(X)) =  X  ''(p(X)) * depends on certain conditions *

Page 14 Inversion Elimination - Conditions - I Eliminated atom must contain all logical variables in parfactors involved. sick(P,D) epidemic(D)

Page 15 Inversion Elimination - Conditions - I Eliminated atom must contain all logical variables in parfactors involved. sick(P,D) epidemic(D) Ok, contains both P and D

Page 16 Inversion Elimination - Conditions - I Eliminated atom must contain all logical variables in parfactors involved. sick(P,D) epidemic(D) Not Ok, missing P sick(P,D)

Page 17 Inversion Elimination - Conditions - I Eliminated atom must contain all logical variables in parfactors involved. q(Y,Z) p(X,Y) No atom can be eliminated

Page 18 Inversion Elimination - Conditions - I … sick(mary, flu) epidemic(flu) sick(mary, rubella) epidemic(rubella) … sick(mary, D) epidemic(D) D  measles Eliminated atom must contain all logical variables - guarantees that subproblems are disjoint.

Page 19 Inversion Elimination - Conditions - II friends(mary,bob) friends(bob,mary) friends(Y,X) friends(X,Y) friends(bob,mary) … X  Y Eliminated RVs must occur in only one instance of parfactor friends(mary,bob) … Inversion Elimination Not Ok

Page 20  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 =  # number of assignments (#)  v1,v2  (v1,v2) #v1,v2 Counting Elimination - A Combinatorial Approach

Page 21 No shared logical variables between atoms, so counting can be done independently  (epidemic(D 1, Region), epidemic(D 2, Region)) Counting Elimination - A Combinatorial Approach

Page 22 Uncovered by Inversion and Counting Eliminating epidemic from   epidemic(Disease1,Region), epidemic(Disease2,Region), donations) No atom with all logical variables, so no Inversion Elimination Shared logical variables, so no Counting Elimination

Page 23 Partial Inversion  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  ’  d ) =  ’  d ) |R| =  ’’  d ) Subsumes Inversion elimination, which is particular case where all logical variables are inverted.

Page 24 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 25 Partial inversion conditions   friends(X,Y), friends(Y,X), sick(X,D), sick(Y,D) ) Cannot partially invert on X,Y because friends(bob,mary) appears in more than one instance of parfactor. But now we do not need all logical variables.

Page 26 Second contribution: Lifted MPE In propositional case, MPE done by factors containing MPE of eliminated variables. AB C D

Page 27 MPE AB D BD  000.3C= C= C= C=1 In propositional case, MPE done by factors containing MPE of eliminated variables.

Page 28 MPE AB B  00.5C=1,D=0 11.4C=1,D=1 In propositional case, MPE done by factors containing MPE of eliminated variables.

Page 29 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 30 MPE  0.9A=0,B=1,C=1,D=1 In propositional case, MPE done by factors containing MPE of eliminated variables.

Page 31 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 32 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 33 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 34 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 35 Conclusions Partial Inversion: More general algorithm, subsumes Inversion elimination Lifted 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 36