Scribd Logo
Upload

Welcome to Scribd!

  • 54 views

    Lec05 Uninformed Search

    Uploaded by

    nayanisateesh2805
    The document discusses uninformed search strategies, which use only information from the problem definition to guide the search. It describes breadth-first search, depth-first search, iterative deepening search, and uniform-cost search. These strategies are evaluated based on completeness, optimality, time complexity and space complexity. Breadth-first search is complete and optimal but has high space complexity, while depth-first search has low space but is not optimal. Uniform-cost search finds lowest-cost solutions when step costs vary.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Lec05 Uninformed Search

    Uploaded by

    nayanisateesh2805
    54 views24 pages
    The document discusses uninformed search strategies, which use only information from the problem definition to guide the search. It describes breadth-first search, depth-first search, iterative deepening search, and uniform-cost search. These strategies are evaluated based on completeness, optimality, time complexity and space complexity. Breadth-first search is complete and optimal but has high space complexity, while depth-first search has low space but is not optimal. Uniform-cost search finds lowest-cost solutions when step costs vary.

    Original Description:

    notes

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document discusses uninformed search strategies, which use only information from the problem definition to guide the search. It describes breadth-first search, depth-first search, iterative deepening search, and uniform-cost search. These strategies are evaluated based on completeness, optimality, time complexity and space complexity. Breadth-first search is complete and optimal but has high space complexity, while depth-first search has low space but is not optimal. Uniform-cost search finds lowest-cost solutions when step costs vary.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    54 views24 pages

    Lec05 Uninformed Search

    Uploaded by

    nayanisateesh2805
    The document discusses uninformed search strategies, which use only information from the problem definition to guide the search. It describes breadth-first search, depth-first search, iterative deepening search, and uniform-cost search. These strategies are evaluated based on completeness, optimality, time complexity and space complexity. Breadth-first search is complete and optimal but has high space complexity, while depth-first search has low space but is not optimal. Uniform-cost search finds lowest-cost solutions when step costs vary.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 24

    Uninformed

     search  strategies  
    •  A  search  strategy  is  defined  by  picking  the  
    order  of  node  expansion  
    •  Uninformed  search  strategies  use  only  the  
    informa:on  available  in  the  problem  defini:on  
    –  Breadth-­‐first  search  
    –  Depth-­‐first  search  
    –  Itera:ve  deepening  search  
    –  Uniform-­‐cost  search  
    Breadth-­‐first  search  
    •  Expand  shallowest  unexpanded  node  
    •  Implementa:on:  fron%er  is  a  FIFO  queue  

    Example  state  space  


    graph  for  a  :ny  search  
    problem  

    Example from P. Abbeel and D. Klein


    Breadth-­‐first  search  
    •  Expansion  order:  
    (S,d,e,p,b,c,e,h,r,q,a,a,  
    h,r,p,q,f,p,q,f,q,c,G)  
    Depth-­‐first  search  
    •  Expand  deepest  unexpanded  node  
    •  Implementa:on:  fron%er  is  a  LIFO  queue  
    Depth-­‐first  search  
    •  Expansion  order:    
    (d,b,a,c,a,e,h,p,q,q,  
    r,f,c,a,G)  
    http://xkcd.com/761/
    Analysis  of  search  strategies  
    •  Strategies  are  evaluated  along  the  following  criteria:  
    –  Completeness:  does  it  always  find  a  solu:on  if  one  exists?  
    –  Op7mality:  does  it  always  find  a  least-­‐cost  solu:on?  
    –  Time  complexity:  number  of  nodes  generated  
    –  Space  complexity:  maximum  number  of  nodes  in  memory  
    •  Time  and  space  complexity  are  measured  in  terms  of    
    –  b:  maximum  branching  factor  of  the  search  tree  
    –  d:  depth  of  the  op:mal  solu:on  
    –  m:  maximum  length  of  any  path  in  the  state  space  (may  be  infinite)  
    Proper:es  of  breadth-­‐first  search  
    •  Complete?    
    Yes  (if  branching  factor  b  is  finite)  
    •  Op7mal?    
    Yes  –  if  cost  =  1  per  step  
    •  Time?    
    Number  of  nodes  in  a  b-­‐ary  tree  of  depth  d:  O(bd)  
    (d  is  the  depth  of  the  op:mal  solu:on)  
    •  Space?    
    O(bd)  

    •  Space  is  the  bigger  problem  (more  than  :me)  


    Proper:es  of  depth-­‐first  search  
    •  Complete?  
    Fails  in  infinite-­‐depth  spaces,  spaces  with  loops  
    Modify  to  avoid  repeated  states  along  path  
    à  complete  in  finite  spaces  
    •  Op7mal?  
    No  –  returns  the  first  solu:on  it  finds  
    •  Time?    
    Could  be  the  :me  to  reach  a  solu:on  at  maximum  depth  m:  O(bm)  
    Terrible  if  m  is  much  larger  than  d  
    But  if  there  are  lots  of  solu:ons,  may  be  much  faster  than  BFS  
    •  Space?    
    O(bm),  i.e.,  linear  space!  
    Itera:ve  deepening  search  
    •  Use  DFS  as  a  subrou:ne  
    1.  Check  the  root  
    2.  Do  a  DFS  searching  for  a  path  of  length  1  
    3.  If  there  is  no  path  of  length  1,  do  a  DFS  
    searching  for  a  path  of  length  2  
    4.  If  there  is  no  path  of  length  2,  do  a  DFS  
    searching  for  a  path  of  length  3…  
    Itera:ve  deepening  search  
    Itera:ve  deepening  search  
    Itera:ve  deepening  search  
    Itera:ve  deepening  search  
    Proper:es  of  itera:ve  deepening  
    search  
    •  Complete?  
    Yes  
    •  Op7mal?  
    Yes,  if  step  cost  =  1  
    •  Time?  
    (d+1)b0  +  d  b1  +  (d-­‐1)b2  +  …  +  bd  =  O(bd)  
    •  Space?  
    O(bd)  
    Search  with  varying  step  costs  

    •  BFS  finds  the  path  with  the  fewest  steps,  but  


    does  not  always  find  the  cheapest  path  
    Uniform-­‐cost  search  
    •  For  each  fron:er  node,  save  the  total  cost  of  
    the  path  from  the  ini:al  state  to  that  node  
    •  Expand  the  fron:er  node  with  the  lowest  path  
    cost  
    •  Implementa:on:  fron%er  is  a  priority  queue  
    ordered  by  path  cost    
    •  Equivalent  to  breadth-­‐first  if  step  costs  all  equal  
    •  Equivalent  to  Dijkstra’s  algorithm  in  general  
    Uniform-­‐cost  search  example  
    Uniform-­‐cost  search  example  
    •  Expansion  order:  
    (S,p,d,b,e,a,r,f,e,G)  
    Another  example  of  uniform-­‐cost  
    search  

    Source: Wikipedia
    Proper:es  of  uniform-­‐cost  search  
    •  Complete?    
    Yes,  if  step  cost  is  greater  than  some  posi:ve  constant  ε  
    (we  don’t  want  infinite  sequences  of  steps  that  have  a  
    finite  total  cost)  
    •  Op7mal?  
    Yes  
    Op:mality  of  uniform-­‐cost  search  
    •  Graph  separa7on  property:  every  path  from  
    the  ini:al  state  to  an  unexplored  state  has  to   start
    pass  through  a  state  on  the  fron:er  
    –  Proved  induc:vely  

    •  Op:mality  of  UCS:  proof  by  contradic:on  


    frontier
    –  Suppose  UCS  terminates  at  goal  state  n  
    with  path  cost  g(n)  but  there  exists   n’’
    another  goal  state  n’  with  g(n’)  <  g(n)  
    –  By  the  graph  separa:on  property,  there   n’
    must  exist  a  node  n”  on  the  fron:er  that  
    is  on  the  op:mal  path  to  n’  
    n
    –  But  because  g(n”)  ≤  g(n’)  <  g(n),    
    n”  should  have  been  expanded  first!  
    Proper:es  of  uniform-­‐cost  search  
    •  Complete?    
    Yes,  if  step  cost  is  greater  than  some  posi:ve  constant  ε  (we  
    don’t  want  infinite  sequences  of  steps  that  have  a  finite  total  
    cost)  
    •  Op7mal?  
    Yes  –  nodes  expanded  in  increasing  order  of  path  cost  
    •  Time?    
    Number  of  nodes  with  path  cost  ≤  cost  of  op:mal  solu:on  (C*),  
    O(bC*/  ε)  
    This  can  be  greater  than  O(bd):  the  search  can  explore  long  
    paths  consis:ng  of  small  steps  before  exploring  shorter  paths  
    consis:ng  of  larger  steps    
    •  Space?    
    O(bC*/  ε)  
    Review:  Uninformed  search  strategies  
    Time   Space  
    Algorithm   Complete?   Op7mal?  
    complexity   complexity  
    Yes   If  all  step     O(bd)   O(bd)  
    BFS   costs  are  equal  

    DFS   No   No   O(bm)   O(bm)  

    If  all  step     O(bd)  


    IDS   Yes   O(bd)  
    costs  are  equal  

    UCS   Yes   Yes   Number  of  nodes  with  g(n)  ≤  C*  

    b:        maximum  branching  factor  of  the  search  tree  


    d:        depth  of  the  op:mal  solu:on  
    m:      maximum  length  of  any  path  in  the  state  space  
    C*:    cost  of  op:mal  solu:on  
    g(n):  cost  of  path  from  start  state  to  node  n  

    You might also like