Symbolic Variable Elimination in Discrete and Continuous Graphical Models. Scott Sanner Ehsan Abbasnejad

Transcription

1 Symbolic Variable Elimination in Discrete and Continuous Graphical Models Scott Sanner Ehsan Abbasnejad

2 Inference for Dynamic Tracking No one previously did this inference exactly in closed-form!

3 Exact Closed-form Continuous Inference Fully Gaussian Most inference including conditional Fully Uniform 1D, n-d hyperrectangular cases General Uniform Piecewise, Asymmetrical Yes, but not a solution you can write on 1 sheet of paper Exact (conditional) inference possible in closed-form?

4 Already Solved? How is inference done in piecewise models? (Adaptively) discretize model: Approximate, O(N D ) Adaptivity is an artform Projective approximation: variational, EP Choose approximating class Often Gaussian good luck! Sampling: Monte Carlo Gibbs May not converge for (near) deterministic distributions Not for unobserved evidence E[ x o=? ] No expressive, exact, closed-form solutions! Yet.

5 What has everyone been missing? Symbolic representations and operations on piecewise functions Continuous variant of work started in (Boutilier, Reiter, Price, IJCAI-01)

6 Piecewise Functions (Cases) z = f(x; y) = Constraint ( (x > 3) ^ (y x) : x + y (x 3) _ (y > x) : x 2 + xy 3 Partition Value Linear constraints and value Linear constraints, constant value Quadratic constraints and value

7 Formal Problem Statement General continuous graphical models represented by piecewise functions (cases) 8 Á >< 1 : f 1 f =.. >: Á k : For variable elimination, need exact closed-form solution inferred via the following piecewise calculus: f k f 1 f 2, f 1 f 2 max( f 1, f 2 ), min( f 1, f 2 ) x f(x) Question: how do we perform these operations in closed-form?

8 Polynomial Case Operations:, ( Á 1 : f 1 Á 2 : f 2 ( Ã 1 : g 1 Ã 2 : g 2 =?

9 Polynomial Case Operations:, ( Á 1 : f 1 Á 2 : f 2 ( Ã 1 : g 1 Ã 2 : g 2 = 8 Á 1 ^ Ã 1 : f 1 + g 1 >< Á 1 ^ Ã 2 : f 1 + g 2 Á 2 ^ Ã 1 : f 2 + g 1 >: Á 2 ^ Ã 2 : f 2 + g 2 Similarly for Polynomials closed under +, * What about max? Max of polynomials is not a polynomial

10 Polynomial Case Operations: max max à ( Á 1 : f 1 Á 2 : f 2 ; ( à 1 : g 1 à 2 : g 2! =?

11 Polynomial Case Operations: max 8 Á 1 ^ à 1 ^ f 1 > g 1 : f 1 Á 1 ^ à 1 ^ f 1 g 1 : g 1 max à ( Á 1 : f 1 Á 2 : f 2 ; ( à 1 : g 1 à 2 : g 2! = Á 1 ^ à 2 ^ f 1 > g 2 : f 1 >< Á 1 ^ à 2 ^ f 1 g 2 : g 2 Á 2 ^ à 1 ^ f 2 > g 1 : f 2 Á 2 ^ à 1 ^ f 2 g 1 : g 1 Á 2 ^ à 2 ^ f 2 > g 2 : f 2 >: Á 2 ^ à 2 ^ f 2 g 2 : g 2 Still a piecewise polynomial! Size blowup? We ll get to that

12 Integration: x x closed for polynomials But how to compute for case? Z x 8 Á >< 1 : f 1.. dx = >: Á k : f k Z x = X i kx [Á i ] f i dx i=1 Z x [Á i ] f i dx Just integrate case partitions, results!

13 Partition Integral 1. Determine integration bounds Z x [Á 1 ] f 1 dx Á 1 := [x > 1] ^ [x > y 1] ^ [x z] ^ [x y + 1] ^ [y > 0] f 1 := x 2 xy LB := ( y 1 > 1 : y 1 y 1 1 : 1 UB := ( z < y + 1 : z z y + 1 : y + 1 What constraints here? independent of x pairwise UB > LB [Á cons ] UB R x = LB f 1 dx UB and LB are symbolic! How to evaluate?

14 Definite Integral Evaluation How to evaluate integral bounds? LB := Z UB x=lb ( y 1 > 1 : y 1 y 1 1 : 1 x 2 xy = 1 3 x3 1 2 x2 y UB := UB LB ( z < y + 1 : z z y + 1 : y + 1 Can do polynomial operations on cases! f 1 UB LB 1 = 3 UB UB UB ª 1 UB UB (y) 2 1 ª 3 LB LB LB ª 1 LB LB (y) 2 Symbolically, exactly evaluated!

15 Exact Graphical Model Inference! (directed and undirected) Can do general probabilistic inference p(x 2 jx 1 ) = R x 3 R x n N k i=1 case i dx n dx 3 R x 2 R x n N k i=1 case i dx n dx 2 Or an exact expectation of any polynomial Z E x»p(xjo) [poly(x)jo] = p(xjo)poly(x)dx poly = Mean, variance, skew, curtosis,, x 2 +y 2 +xy x All computed by Symbolic Variable Elimination (SVE)

16 Voila: Closed-form Exact Inference via SVE!

17 Bayesian Robotics D x 1,, x n D: true distance to wall x 1,, x n : measurements want: E[D x 1,,x n-1, x n =?]

18 Bayesian Robotics: Inference Examples Posterior distance shown in graph Expected distance calculation shown below

19 Extra: A New Conjugate Prior for Bayesian Inference General Bayesian Inference p( ~ µjd n+1 ) / p(d n+1 j ~ µ)p( ~ µjd n ) Prior and likelihood are case statements then posterior is a case statement

20 Example: Bayesian Preference Learning for Linear Utilities If prior is uniform over weights and each preference likelihood linearly constrains weights posterior is uniform over convex polytope Inference is equivalent All case statements! to finding volume of a convex polytope!

21 Computational Complexity? In theory for SVE on graphical models Best-case complexity (#operations) Worst-case complexity is O(exp(#operations)) Not explicitly tree-width dependent! But worse: integral may invoke 100 s of operations! Fortunately data structures mitigate worstcase complexity

22 BDD / ADDs Quick Introduction

23 Function Representation (Tables) How to represent functions: B n R? How about a fully enumerated table OK, but can we be more compact? a b c F(a,b,c)

24 Function Representation (Trees) How about a tree? Sure, can simplify. a b c F(a,b,c) Context-specific independence! c a b c

25 Function Representation (ADDs) Why not a directed acyclic graph (DAG)? a b c F(a,b,c) a c b Algebraic Decision Diagram (ADD) Think of BDDs as {0,1} subset of ADD range

26 Binary Operations (ADDs) Why do we order variable tests? Enables us to do efficient binary operations a b a c Result: ADD operations can avoid state enumeration a b c 0 2 c

27 Case XADD XADD = continuous variable extension of algebraic decision diagram Efficient XADD data structure for cases strict ordering of atomic inequality tests compact, minimal case representation efficient case operations

28 Case XADD 8 x 1 + k > 100 ^ x 2 + k > 100 : 0 x 1 + k > 100 ^ x 2 + k 100 : x 2 >< x 1 + k 100 ^ x 2 + k > 100 : x 1 V = x 1 + x 2 + k > 100 ^ x 1 + k 100 ^ x 2 + k 100 ^ x 2 > x 1 : x 2 x 1 + x 2 + k > 100 ^ x 1 + k 100 ^ x 2 + k 100 ^ x 2 x 1 : x 1 x 1 + x 2 + k 100 : x 1 + x 2 >:.. *With non-trivial extensions over ADD, can reduce to a minimal canonical form!

29 Compactness of (X)ADDs Linear in number of decisions i Case version has exponential number of partitions!

30 XADD Maximization y > 0 max( y > 0, x > 0 ) = x > 0 x > 0 x y y x x > y May introduce new decision tests x y Operations exploit structure: O( f g )

31 Maintaining XADD Orderings Max may get decisions out of order Decision ordering (root leaf) x > y y > 0 x > 0 max( y > 0, x > 0 ) = x y y Newly introduced node is out of order! x y > 0 x > 0 x > 0 x > y x y

32 Correcting XADD Ordering Obtain ordered XADD from unordered XADD key idea: binary operations maintain orderings z is out of order z result will have z in order! ID 1 z 1 0 ID 0 z 0 1 ID 1 ID 0 Inductively assume ID 1 and ID 0 are ordered. All operands ordered, so applying, produces ordered result!

33 Maintaining Minimality y > 0 y > 0 x > 0 x > 0 x + y < 0 y x y x + y x + y Node unreachable x + y < 0 always false if x > 0 & y > 0 If linear, can detect with feasibility checker of LP solver & prune More subtle prunings as well.

34 XADD Makes Possible all Previous Inference Could not even do a single integral or maximization without it!

35 Recap Defined a calculus for piecewise functions f 1 f 2, f 1 f 2 max( f 1, f 2 ), min( f 1, f 2 ) x f(x) See Zamani & Sanner (AAAI-12) for max x f(x), min x f(x) Defined XADD to efficiently compute with cases Makes possible Closed-form inference in continuous graphical models Exact (conditional) expectations of any polynomial New paradigms for Bayesian inference

36 Piecewise Calculus + XADD = Expressive, Exact, Closed-form Inference in Discrete & Continuous Variable Graphical Models Thank you! Questions?