dynamic programming
play

Dynamic Programming Part 2 Algorithm Theory WS 2012/13 Fabian Kuhn - PowerPoint PPT Presentation

Chapter 3 Dynamic Programming Part 2 Algorithm Theory WS 2012/13 Fabian Kuhn Dynamic Programming Memoization for increasing the efficiency of a recursive solution: Only the first time a sub problem is encountered, its solution is


  1. Chapter 3 Dynamic Programming Part 2 Algorithm Theory WS 2012/13 Fabian Kuhn

  2. Dynamic Programming „Memoization“ for increasing the efficiency of a recursive solution: • Only the first time a sub ‐ problem is encountered, its solution is computed and then stored in a table. Each subsequent time that the subproblem is encountered, the value stored in the table is simply looked up and returned (without repeated computation!). • Computing the solution : For each sub ‐ problem, store how the value is obtained (according to which recursive rule). Algorithm Theory, WS 2012/13 Fabian Kuhn 2

  3. Dynamic Programming Dynamic programming / memoization can be applied if • Optimal solution contains optimal solutions to sub ‐ problems (recursive structure) • Number of sub ‐ problems that need to be considered is small Algorithm Theory, WS 2012/13 Fabian Kuhn 3

  4. String Matching Problems Edit distance: • For two given strings � and � , efficiently compute the edit distance ���, �� (# edit operations to transform � into � ) as well as a minimum sequence of edit operations that transform � into � . • Example: mathematician  multiplication: m a t h e m a t i c i a n u i p l o l i c Algorithm Theory, WS 2012/13 Fabian Kuhn 4

  5. String Matching Problems Edit distance ���, �� (between strings � and ��: m a – t h e m - - a t i c i a n m u l t i p l i c a t i o - - n Approximate string matching: For a given text T , a pattern P and a distance d , find all substrings �′ of � with ���, �′�  � . Sequence alignment: Find optimal alignments of DNA / RNA / ... sequences. G A G C A - C T T G G A T T C T C G G - - - C A C G T G G - A - A C T - - - Algorithm Theory, WS 2012/13 Fabian Kuhn 5

  6. Edit Distance Given: Two strings � � � � � � … � � and � � � � � � … � � Goal: Determine the minimum number ���, �� of edit operations required to transform � into � Edit operations: a) Replace a character from string � by a character from � b) Delete a character from string � c) Insert a character from string � into � m a – t h e m - - a t i c i a n m u l t i p l i c a t i o - - n Algorithm Theory, WS 2012/13 Fabian Kuhn 6

  7. Edit Distance – Cost Model • Cost for replacing character � by � : � �, � � � • Capture insert, delete by allowing � � � or � � � : – Cost for deleting character � : ���, �� – Cost for inserting character � : ���, �� • Triangle inequality : � �, � � � �, � � � �, �  each character is changed at most once! • Unit cost model : � �, � � �1, if � � � 0, if � � � Algorithm Theory, WS 2012/13 Fabian Kuhn 7

  8. Recursive Structure • Optimal “alignment” of strings (unit cost model) and abbagflrgikacc : bbcadfagikccm - b b c a g f a – g i k - c c m a b b – a d f l r g i k a c c – • Consists of optimal “alignments” of sub ‐ strings, e.g.: -bbcagfa –gik-ccm and abb-adfl rgikacc- • Edit distance between � �,� � � � … � � and � �,� � � � … � � : � �, � � min �,ℓ � � �,� , � �,ℓ � � � ���,� , � ℓ��,� Algorithm Theory, WS 2012/13 Fabian Kuhn 8

  9. Computation of the Edit Distance Let � � ≔ � � … � � , � ℓ ≔ � � … � ℓ , and � �,ℓ ≔ � � � , � ℓ Algorithm Theory, WS 2012/13 Fabian Kuhn 9

  10. Computation of the Edit Distance Three ways of ending an “alignment” between � � and � ℓ : � � is replaced by � ℓ : 1. � �,ℓ � � ���,ℓ�� � � � � , � ℓ � � is deleted: 2. � �,ℓ � � ���,ℓ � � � � , � 3. � ℓ is inserted: � �,ℓ � � �,ℓ�� � � �, � ℓ Algorithm Theory, WS 2012/13 Fabian Kuhn 10

  11. Computing the Edit Distance • Recurrence relation (for �, ℓ � 1 ) � ���,ℓ�� � � � � , � ℓ � ���,ℓ�� � 1 � ���,ℓ � 1 � ���,ℓ � � � � , � � �,ℓ � min � min � �,ℓ�� � 1 � �,ℓ�� � � �, � ℓ unit cost model • Need to compute � �,� for all 0 � � � � , 0 � � � ℓ : � ���,ℓ�� � ���,ℓ�� � ���,ℓ � ���,ℓ �� �� � �,ℓ�� � �,ℓ�� � �,ℓ � �,ℓ �� Algorithm Theory, WS 2012/13 Fabian Kuhn 11

  12. Recurrence Relation for the Edit Distance Base cases: � �,� � � �, � � � � �,� � � �, � � � � �,��� � � �, � � � �,� � � � � , � � � ���,� � � � � , � Recurrence relation: � ���,ℓ�� � � � � , � ℓ � ���,ℓ � � � � , � � �,� � ��� � �,ℓ�� � � �, � ℓ Algorithm Theory, WS 2012/13 Fabian Kuhn 12

  13. Order of solving the subproblems � 1 � 2 � 3 � 4 … � � � 1 � 2 � � � ���,��� � �,��� � ���,� � �,� Algorithm Theory, WS 2012/13 Fabian Kuhn 13

  14. Algorithm for Computing the Edit Distance Algorithm Edit ‐ Distance 2 strings � � � � … � � and � � � � … � � Input: Output: matrix � � � �� 1 � 0,0 ≔ 0 ; 2 for � ≔ 1 to � do � �, 0 ≔ � ; 3 for � ≔ 1 to � do � 0, � ≔ � ; 4 fo r � ≔ 1 to � do for � ≔ 1 to � do 5 � � � 1, � � 1 � �, � � 1 � 1 � �, � ≔ min 6 ; � � � 1, � � 1 � � � � , � � Algorithm Theory, WS 2012/13 Fabian Kuhn 14

  15. Example Algorithm Theory, WS 2012/13 Fabian Kuhn 15

  16. Computing the Edit Operations Algorithm Edit ‐ Operations ��, �� Input: matrix � (already computed) Output: list of edit operations 1 if � � 0 and � � 0 then return empty list 2 if � � 0 and � �, � � � � � 1, � � 1 then return Edit ‐ Operations �� � 1, �� ∘ „delete � � “ 3 4 else if � � 0 and � �, � � � �, � � 1 � 1 then 5 return Edit ‐ Operations ��, � � 1� ∘ „insert � � “ 6 else // � �, � � � � � 1, � � 1 � ��� � , � � � if � � � � � then return Edit ‐ Operations �� � 1, � � 1� 7 8 else return Edit ‐ Operations �� � 1, � � 1� ∘ „replace � � by � � “ Initial call: Edit ‐ Operations ( m,n ) Algorithm Theory, WS 2012/13 Fabian Kuhn 16

  17. Edit Operations Algorithm Theory, WS 2012/13 Fabian Kuhn 17

  18. Edit Distance: Summary • Edit distance between two strings of length � and � can be computed in � �� time. • Obtain the edit operations: – for each cell, store which rule(s) apply to fill the cell – track path backwards from cell ��, �� – can also be used to get all optimal “alignments” • Unit cost model: – interesting special case – each edit operation costs 1 Algorithm Theory, WS 2012/13 Fabian Kuhn 18

  19. Approximate String Matching Given: strings � � � � � � … � � (text) and � � � � � � … � � (pattern). Goal: Find an interval ��, �� , 1 � � � � � � such that the sub ‐ string � �,� ≔ � � … � � is the one with highest similarity to the pattern � : arg min � � �,� , � ������� � � � � � Algorithm Theory, WS 2012/13 Fabian Kuhn 19

  20. Approximate String Matching Naive Solution: for all 1 � � � � � � do compute ��� �,� , �� choose the minimum Algorithm Theory, WS 2012/13 Fabian Kuhn 20

  21. Approximate String Matching A related problem: • For each position � in the text and each position � in the pattern compute the minimum edit distance ���, �� between � � � � � … � � and any substring � �,� of � that ends at position � . � � � � � � � � … � � � ���, �� Algorithm Theory, WS 2012/13 Fabian Kuhn 21

  22. Approximate String Matching Three ways of ending optimal alignment between � � and � � : � � is replaced by � � : 1. � �,� � � ���,��� � � � � , � � � � is deleted: 2. � �,� � � ���,� � � � � , � � � is inserted: 3. � �,� � � �,��� � � �, � � Algorithm Theory, WS 2012/13 Fabian Kuhn 22

  23. Approximate String Matching Recurrence relation (unit cost model): � ���,��� � � � ���,� � � � �,� � ��� � �,��� � � Base cases: � �,� � � � �,� � � � �,� � � Algorithm Theory, WS 2012/13 Fabian Kuhn 23

  24. Example 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 1 1 1 0 1 1 1 1 1 2 1 1 2 2 2 1 1 2 2 2 2 3 2 2 2 3 3 2 2 2 3 3 3 4 3 3 2 3 4 3 3 2 3 4 5 5 4 4 3 3 4 4 4 3 2 3 4 Algorithm Theory, WS 2012/13 Fabian Kuhn 24

  25. Approximate String Matching • Optimal matching consists of optimal sub ‐ matchings • Optimal matching can be computed in ����� time • Get matching(s): – Start from minimum entry/entries in bottom row – Follow path(s) to top row • Algorithm to compute ���, �� identical to edit distance algorithm, except for the initialization of ���, 0� Algorithm Theory, WS 2012/13 Fabian Kuhn 25

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend