AI Lecture 3 (solving problems by searching)

CSE 412: Artificial IntelligenceCSE 412: Artificial Intelligence
Fall 2018Fall 2018
Topic – 3: Solving ProblemsTopic – 3: Solving Problems
by Searchingby Searching
Tajim Md. Niamat Ullah Akhund
Lecturer
Department of Computer Science and Engineering
Daffodil International University
Email: tajim.cse@diu.edu.bd

Topic ContentsTopic Contents
Problem Solving AgentsProblem Solving Agents
Problem Formulation and Search SpacesProblem Formulation and Search Spaces
Example ProblemsExample Problems
Tree Search AlgorithmTree Search Algorithm
 Breadth-First SearchBreadth-First Search
 Uniform-Cost SearchUniform-Cost Search
 Depth-First SearchDepth-First Search
 Depth Limited SearchDepth Limited Search
 Iteratively Deepening SearchIteratively Deepening Search
 Bidirectional SearchBidirectional Search 2

IntroductionIntroduction
In this topic, we see how an agent can find
a sequence of actions that achieves its
goals, when no single action will do.
3
Majidur RahmanMajidur Rahman

Problem-SolvingProblem-Solving AgentAgent
Four general steps in problem solving:
• Goal formulation
– What are the successful world states
• Problem formulation
– What actions and states to consider given the goal
• Search
– Examine different possible sequences of actions that
lead to states of known value and then choose the
best sequence
• Execute
– Perform actions on the basis of the solution
4

Example: RomaniaExample: Romania
5

Example: RomaniaExample: Romania
On holiday the agent is in Romania visiting in
Arad. Flight leaves tomorrow from Bucharest.
• Formulate goal
– Be in Bucharest
• Formulate problem
– States: various cities
– Actions: drive between cities
• Find solution
– Sequence of cities; e.g. Arad, Sibiu, Fagaras,
Bucharest,…
6

Problem TypeProblem Type
Given how we have defined the problem:
• Environment is fully observable
• Environment is deterministic
• Environment is sequential
• Environment is static
• Environment is discrete
• Environment is single-agent
7

Problem FormulationProblem Formulation
A problem is defined by:
 An initial state, e.g. In(Arad)
 A successor function S(x) = set of action-state pairs
• S(In(Arad)) = {(Go(Sibiu), In(Sibiu)), (Go(Zerind), In(Zerind)),
(Go(Timisoara), In(Timisoara))}
• initial state + successor function = state space
 Goal test
• Explicit, e.g. x=‘In(Bucharest)’
 Path cost (assume additive)
• e.g. sum of distances, number of actions executed, …
A solution is a sequence of actions from initial to
goal state
 the optimal solution has the lowest path cost
8

Vacuum WorldVacuum World
State space of the vacuum world
9

Example ProblemsExample Problems
The problem-solving approach has been applied to
a vast array of task environments.
Some of the best known are listed here,
distinguishing between toy and real-world problems.
A toy problem is intended to illustrate or exercise
various problem-solving methods.
It is usable by different researchers to compare the
performance of algorithms.
 A real-world problem is one whose solutions
people actually care about.
We will mainly focus on toy problems in this topic.
10

Toy Problems: 8-PuzzleToy Problems: 8-Puzzle
 States: location of each tile plus blank
 Initial state: Any state can be initial
 Actions: Move blank {Left, Right, Up, Down}
 Goal test: Check whether goal configuration is reached
 Path cost: Number of actions to reach goal
11

Toy Problems: 8-QueensToy Problems: 8-Queens
Place eight queens on a chessboard such that
no queen attacks any other.
Two kinds of problem formulation:
• Complete-state formulation
• Incremental formulation
12

Complete-state formulation
• States: Any arrangement of 0 to 8 queens on the board
• Initial state: No queens
• Actions: Add queen to any empty square
• Goal test: 8 queens on board and none attacked
• Path cost: N/A
64.63.62…..57 ≈ 3 x 1014
possible sequences to investigate
13Tajim Md. Niamat Ullah AkhundTajim Md. Niamat Ullah Akhund

Incremental formulation
• States: n (0 to 8) queens on the board, one per column in
the n leftmost columns with no queen attacking any other
• Actions: Add queen in leftmost empty column such that
the queen is not attacking any other queen
• 2057 possible sequences to investigate; solutions are
easy to find
But with 100 queens the problem is much harder
14

Real-World ProblemsReal-World Problems
 Route-finding problems
 Touring problems
 Traveling Salesman problem
 VLSI layout problem
 Robot navigation
 Automatic assembly sequencing
 Internet searching
15

Basic Search AlgorithmsBasic Search Algorithms
How do we find the solutions of previous
problems?
• Search the state space
– State space can be represented by a search tree
– Root of the tree is the initial state
– Children generated through successor function
• In general we may have a search graph rather
than tree (same state can be reached through
multiple paths)
16

Simple Tree Search ExampleSimple Tree Search Example
17

18

19

State Space vs. Search TreeState Space vs. Search Tree
• A state corresponds
to a configuration of
the world
• A node is a data
structure in a search
tree
– e.g. node = <state,
parent-node, action,
path-cost, depth>
20

Search StrategiesSearch Strategies
 A search strategy is defined by picking the order of node
expansion
 Problem-solving performance is measured in four ways:
• Completeness: Is a solution found if one exists?
• Optimality: Does the strategy find the optimal solution?
• Time Complexity: How long does it take to find a solution?
• Space Complexity: How much memory is needed to perform
the search?
 Time and space complexity are measured in terms of
problem difficulty defined by:
• b - branching factor of the search tree
• d - depth of the shallowest goal node
• m - maximum length of any path in the state space
21

Uninformed Search StrategiesUninformed Search Strategies
• Uninformed search (or blind search)
– Strategies have no additional information about states beyond
that provided in the problem definition
• Informed (or heuristic) search
– Search strategies know whether one state is more promising
than another
• Uninformed strategies (defined by order in which nodes
are expanded):
– Breadth-first search
– Uniform-cost search
– Depth-first search
– Depth-limited search
– Iterative deepening search
– Bidirectional search
22

Breadth-First SearchBreadth-First Search
• Expand shallowest unexpanded node
• Fringe is the collection of nodes that have been generated
but not yet expanded.
• The set of all leaf nodes available for expansion at any
given point is called the frontier.
• Implementation: fringe/frontier is a FIFO queue
23

24

25

26

• Completeness: is a solution always found
if one exists?
– YES
• If shallowest goal node is at some finite depth d
• If branching factor b is finite
• BF search is optimal if the path cost is a
non-decreasing function of the depth of
the node
27

• Time complexity: Assume a state space where
every state has b successors
– Assume solution is at depth d
– Worst case: expand all but the last node at depth d
– Total number of nodes generated:
– b + b2
+ b3
+ ...+ bd
+ (bd +1
− b) = O(bd +1
)
• Space complexity: every node generated must
remain in memory, so same as time complexity
28

• Memory requirements are a bigger problem than
execution time.
29

Uniform-Cost SearchUniform-Cost Search
• Extension of BF-search:
– Expand node with lowest path cost
• Implementation: fringe = queue ordered by
path cost
• UC-search is the same as BF-search
when all step costs are equal
30

 Let us consider the following search tree, where A
and G are the initial and goal node, respectively.
31

32

33

34

35

36

 Let us consider the another search tree, where A
and G are the initial and goal node, respectively.
37

38

39

40

41

• Completeness:
– YES, if step-cost > some small positive constant ε
• Optimality:
– nodes expanded in order of increasing path cost
– YES, if complete
• Time and space complexity:
– Uniform-cost search is guided by path costs rather than
depths, so its complexity cannot easily be characterized
in terms of b and d.
– Instead, assume C* is the cost of the optimal solution
– Assume that every action costs at least ε
– Worst-case: O(bC*/ε
)
42

Depth-First SearchDepth-First Search
• Expand deepest unexpanded node
• Implementation: fringe is a LIFO queue (= stack)
43

44

45

46

47

48

49

50

51

52

53

54

DF-Search: EvaluationDF-Search: Evaluation
• Completeness:
– Is a solution always found if one exists?
– No (unless search space is finite and no loops
are possible)
• Optimality:
– Is the least-cost solution always found?
– No
55

DF-Search: EvaluationDF-Search: Evaluation
• Time complexity: O(bm
)
• In general, time is terrible if m (maximal
depth) is much larger than d (depth of
shallowest solution)
– But if there exist many solutions then faster
than BF-search
• Space complexity: O(bm)
– Backtracking search uses even less memory
(one successor instead of all b)
56

Depth-Limited SearchDepth-Limited Search
• DF-search with depth limit l
– i.e. nodes at depth l have no successors
– Problem knowledge can be used
• Solves the infinite-path problem
• If l < d then incompleteness results
• Time complexity: O(bl
)
• Space complexity: O(bl)
57

Depth-Limited Search withDepth-Limited Search with ll = 2= 2
58

59

60

61

Iterative Deepening SearchIterative Deepening Search
• A general strategy to find best depth limit l
– Goal is found at depth d, the depth of the
shallowest goal-node
– Often used in combination with DF-search
• Combines benefits of DF- and BF-search
62

ID-Search: ExampleID-Search: Example
63

64

65Majidur RahmanMajidur Rahman

Bidirectional SearchBidirectional Search
Two simultaneous searches run from start and goal.
– Motivation:
– One forward from the initial state
– Other backward from goal
– Stops when the two searches meet in the middle
• Check whether the node belongs to the other fringe before
expansion
67
bd/2
+bd/2
≠bd

Time complexity: O(bd/2
)
Space complexity: O(bd/2
)
– This space complexity is the most significant
weakness of bidirectional search.
Completeness and Optimality: Yes
– If step costs are uniform
– If both searches are breadth-first search.

The reduction in time complexity makes
bidirectional search attractive, but how do we
search backwards?
The predecessor of each node should be
efficiently computable.
– When actions are easily reversible.

Summary of AlgorithmsSummary of Algorithms
70

********** ********** If You Need Me ********** **********
Mail: tajim.cse@diu.edu.bd
Website: https://www.tajimiitju.blogspot.com
ORCID: https://orcid.org/0000-0002-2834-1507
LinkedIn: https://www.linkedin.com/in/tajimiitju
ResearchGate: https://www.researchgate.net/profile/Tajim_Md_Niamat_Ullah_Akhund
YouTube: https://www.youtube.com/tajimiitju?sub_confirmation=1
SlideShare: https://www.slideshare.net/TajimMdNiamatUllahAk
Facebook: https://www.facebook.com/tajim.mohammad
GitHub: https://github.com/tajimiitju
Google+: https://plus.google.com/+tajimiitju
Gmail: tajim.mohammad.3@gmail.com
Twitter: https://twitter.com/Tajim53
Thank you
Tajim Md. Niamat Ullah AkhundTajim Md. Niamat Ullah Akhund

AI Lecture 3 (solving problems by searching)

In this document

More Related Content

What's hot

Similar to AI Lecture 3 (solving problems by searching)

More from Tajim Md. Niamat Ullah Akhund

Recently uploaded

AI Lecture 3 (solving problems by searching)