1 / 68

Hybrid of search and inference: time-space tradeoffs chapter 10

Hybrid of search and inference: time-space tradeoffs chapter 10. ICS-275 Spring 2007. Chapter 10: Hybrids of Search and Inference. Interleaving conditioning and elimination Cutset decomposition AND/OR cutset decomposition Super-bucket and super-clusters

bayle
Download Presentation

Hybrid of search and inference: time-space tradeoffs chapter 10

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Hybrid of search and inference: time-space tradeoffschapter 10 ICS-275 Spring 2007

  2. Chapter 10:Hybrids of Search and Inference • Interleaving conditioning and elimination • Cutset decomposition • AND/OR cutset decomposition • Super-bucket and super-clusters • Approximating conditioning in a hybrid ICS-275

  3. Satisfiability: Inference vs search Search = O(exp(n)) Search = exp(dfs-height) ICS-275

  4. DR versus DPLL: Complementary Properties Uniform random 3-CNFs (large induced width) (k,m)-tree 3-CNFs (bounded induced width) ICS-275

  5. Outline; Road Map Tasks Methods ICS-275

  6. The effect of Conditioning X1 X2 X3 X4 X5 ICS-275

  7. The effect of Conditioning • Select a variable X1 X2 X3 X4 X5 ICS-275

  8. The effect of Conditioning X1 X2 X3 X4 X5 …... X1  a X1  c X1  b X2 X2 X2 X3 X3 X3 …... X4 X5 X4 X5 X4 X5 ICS-275

  9. The effect of Conditioning X1 X2 X3 Condition on X2, and BE or BE(w=2) now. X4 X5 …... X1  a X1  c X1  b X2 X2 X2 X3 X3 X3 …... X4 X5 X4 X5 X4 X5 ICS-275

  10. The effect of Conditioning X1 X2 General principle: Condition until tractable X3 X4 X5 …... X1  a X1  c X1  b X2 X2 X2 X3 X3 X3 …... X4 X5 X4 X5 X4 X5 ICS-275

  11. Interleaving Cond and Elim (VE+C) ICS-275

  12. Interleaving Cond and Elim ICS-275

  13. Interleaving Cond and Elim ICS-275

  14. Interleaving Cond and Elim ICS-275

  15. Interleaving Cond and Elim ICS-275

  16. Interleaving Cond and Elim ICS-275

  17. Interleaving Cond and Elim ... ... ICS-275

  18. Interleaving Cond and Elim ... ... ICS-275

  19. Interleaving Cond and Elim ... ... ICS-275

  20. Interleaving Cond and Elim ... ... ICS-275

  21. Interleaving Cond and Elim ... ... ICS-275

  22. Interleaving Cond and Elim ... ... ICS-275

  23. Interleaving Cond and Elim ... ... ICS-275

  24. Interleaving Cond and Elim ... ... ICS-275

  25. Interleaving Cond and Elim ... ... ICS-275

  26. ICS-275

  27. DCDR(b): empirical results ICS-275

  28. Road Map • Interleaving conditioning and elimination • Cutset decomposition • W-cutset, cycle-cutset • AND/OR w-cutset • Super-bucket and super-clusters ICS-275

  29. H H H G G G N N N H H G G D D D N N F F F M M M D D A A A A J J J F F M M O O O J J E E E O O E E C C C P P P B B B C C P P B B K K K L L L K K L L B H H H G G G N N N D D D C F F M M F M J J J O O O E E E C C P P P K K L L K L Cycle cutset Cycle cutset = {A,B,C} ICS-275

  30. E M D Graph Coloring problem L C A B K H F G J A=yellow A=green E M E M D D L L C C B B K K H H F F G G J J Conditioning and cycle-cutset • Inference may require too much memory • Condition (guessing) on some of the variables ICS-275

  31. E M D Graph Coloring problem L C A B K H F G J A=yellow A=green B=red B=blue B=green B=red B=blue B=yellow E E E E E E M M M M M M D D D D D D L L L L L L C C C C C C K K K K K K H H H H H H F F F F F F G G G G G G J J J J J J Conditioning • Inference may require too much memory • Condition on some of the variables ICS-275

  32. The cycle-cutset scheme: condition until a tree ICS-275

  33. Time-space tradeoff Theorem: The w-cutset scheme yields space complexity exp(w) and time complexity exp(w+c_w), where c_w is the size of the w-cutset. As w decreases, c_w increases. Thecycle-cutset decomposition is linear space and Has time complexity of ICS-275

  34. Elim-cond(w) or w-cutset scheme • Idea: runs backtracking search on the w-cutset variables and bucket-elimination on the remaining variables. • Input: A constraint network R = (X,D,C), Y a w-cutset, d an ordering that starts with Y whose adjusted induced-width, along d, is bounded by w, Z = X-Y. • Output: A consistent assignment, if there is one. • 1. while {y}  next partial solution of Y found by backtracking, do • a) z  solution found by adaptive-consistency(R_y). • B) if z is not false, return solution (y,z). • 2. endwhile. • return: the problem has no solutions. ICS-275

  35. Finding a w-cutset • Verifying a w-cutset can be done in polynomial time • A simple greedy: use a good induced-width ordering and starting at the top add to the w-cutset any variable with more than w parents. • Alternative: generate a tree-decomposition and select a w-cutset that reduce each cluster below w (Bidyuk and Dechter UAI2005). ICS-275

  36. Treewidth equals cycle cutset G D G A D A B F C C E L K L K H M M J J N N treewidth = cycle cutset = 4 ICS-275

  37. Treewidth smaller than cycle cutset G D G B D A B F F C C E E K H L J treewidth = 2 cycle cutset = 5 ICS-275

  38. Time-Space complexity of of w-cutset • Space: O(exp(w)) • W-cutset: a set that when removed the induced-width is w. • c(w): size of w-cutset. • Time: O(exp(w+c(w))) on OR space ICS-275

  39. Time vs space • Random Graphs (50 nodes, 200 edges, average degree 8, w*23) 60 Branch and bound 50 40 Bucket elimination W+c(w) 30 20 10 0 2 5 8 -1 11 14 17 20 23 26 29 32 w ICS-275

  40. DCDR(b): empirical results(Rish and Dechter 2000) ICS-275

  41. Empirical evaluation of w-cutset • Cycle-cutset for CSPs (dechter 1990) • Alternating w-cutset for cnfs (Rish and Dechter 2000) • Alternating w-cutset for Constraint optimization (Larrosa and Dechter, 2002) ICS-275

  42. The idea of super-buckets Larger super-buckets (cliques) =>more time but less space • Complexity: • Time: exponential in clique (super-bucket) size • Space: exponential in separator size ICS-275

  43. Example ICS-275

  44. Super-Bucket Elimination, SBE(k) • Eliminate sets of variables such that: • individual eliminations are too costly in space (namely, each variable in the set has degree larger than k) • the join degree is lower than k ICS-275

  45. Sep-based time-space tradeoff • Let T be a tree-decomposition of hypergraph H. Let s_0,s_1,...,s_n be the sizes of the separators in T, listed in strictly descending order. With each separator size s_i we associate a secondary tree decomposition T_i, generated by combining adjacent nodes whose separator sizes are strictly greater than s_i. • We denote by r_i the largest set of variables in any cluster of T_i. • Note that as s_i decreases, r_i increase. • Theorem: The complexity of CTE when applied to each T_i is O( n exp(r_i)) time, and O( n exp(s_i)) space. ICS-275

  46. Complexity • Theorem: If R = (X,D,C) is a constraint network whose constraint graph has non-separable components of at most size r, then the super-bucket elimination algorithm, whose buckets are the non-separable components, is time exponential O(n exp(r)) and is linear in space. ICS-275

  47. Hybrids of hybrids • hybrid(b_1,b_2): • First, a tree-decomposition having separators bounded by b_1 is created, followed by application of the CTE algorithm, but each clique is processed by elim-cond(b_2). If c^*_{b_2} is the size of the maximum b_2-cutset in each clique of the b_1-tree-decomposition, the algorithm is space exponential in b_1 but time exponential in c^*_{b_2}. • Special cases: • hybrid(b_1,1): Applies cycle-cutset in each clique. • b_1 = b_2. For b=1, hybrid(1,1) is the non-separable components utilizing the cycle-cutset in each component. • The space complexity of this algorithm is linear but its time complexity can be much better than the cycle-cutsets cheme or the non-separable component approach alone. ICS-275

  48. Case study: combinatorial circuits: benchmark used for fault diagnosis and testing community Problem: Given a circuit and its unexpected output, identify faulty components. The problem can be modeled as a constraint optimization problem and solved by bucket elimination. ICS-275

  49. Case study: C432 Join-tree of c432 Seperator size is 23 A circuit’s primal graph For every gate we connect inputs and outputs ICS-275

  50. Join-tree of C3540 (1719 vars)max sep size 89 ICS-275

More Related