Artificial Intelligence

INFORMED SEARCH
ENG P KADEBU
 Informed Search

Informed Search – a strategy that uses

problem-specific knowledge beyond the definition
of the problem itself

Best-first search
Greedy best-first search
A* search
 Review: Tree search
function Tree-SEARCH(problem,fringe) return a solution or failure
closed  an empty set
fringe  INSERT(MAKE-NODE(INITIAL-STATE[problem]), fringe))
loop do
if EMPTY?(fringe) then return failure
node  REMOVE-FIRST(fringe)
if GOAL-TEST[problem] applied to STATE[node] succeeds
then return SOLUTION(node)
if STATE[node] is not in closed then
add STATE[node] to closed
fringe  INSERT-ALL(EXPAND(node, problem), fringe)

 A search strategy is defined by picking the order of node expansion

 Heuristic

Heuristic means “rule of thumb”.

To quote Judea Pearl, “Heuristics are criteria,
methods or principles for deciding which among
several alternative courses of action promises to be
the most effective in order to achieve some goal”.
In heuristic search or informed search, heuristics are
used to identify the most promising search path.
 Heuristic Function

A heuristic function at a node n is an estimate of the

optimum cost from the current node to a goal. It is
denoted by h(n).

h(n) = estimated cost of the cheapest path from node n to

a goal node

Example 1: We want a path from Kolkata to Guwahati.

Heuristic for Guwahati may be straight-line distance
between Kolkata and Guwahati h(Kolkata) =
euclideanDistance(Kolkata, Guwahati)
 Best-first search

 Idea: use an evaluation function f(n) for each node

 estimate of "desirability"
 Expand most desirable unexpanded node

 Implementation:
Order the nodes in fringe in decreasing order of desirability
 Special cases:
 greedy best-first search
 A* search

Best-first Search Algorithm
Romania with step costs in km
 Greedy best-first search

Evaluation function f(n) = h(n) (heuristic)

= estimate of cost from n to goal
It estimates the distance remaining to a goal.

e.g., hSLD(n) = straight-line distance from n to Bucharest

Greedy best-first search expands the node that appears to
be closest to goal

Greedy algorithms often perform very well.

They tend to find good solutions quickly, although not
always optimal ones.
Greedy best-first search example
Greedy best-first search example
Greedy best-first search example
Greedy best-first search example
 Properties of greedy best-first search

Complete? No – can get stuck in loops, e.g., Iasi 

Neamt  Iasi  Neamt 
Time? O(bm), but a good heuristic can give
dramatic improvement
Space? O(bm) -- keeps all nodes in memory
Optimal? No
 A* search Algorithm

Idea: avoid expanding paths that are already

expensive
Evaluation function f(n) = g(n) + h(n)
g(n) = cost so far to reach n
h(n) = estimated cost from n to goal
f(n) = estimated total cost of path through n to goal
A* search
A* search example
A* search example
A* search example
A* search example
A* search example
A* search example
Example
Solution
 Admissible heuristics

 A heuristic h(n) is admissible if for every node n,

h(n) ≤ h*(n), where h*(n) is the true cost to reach the goal
state from n.
 An admissible heuristic never overestimates the cost to
reach the goal, i.e., it is optimistic
 Example: hSLD(n) (never overestimates the actual road
distance)

 Theorem: If h(n) is admissible, A* using TREE-SEARCH is

optimal
 Optimality of A* (proof)

 Suppose some suboptimal goal G2 has been generated and is in the

fringe. Let n be an unexpanded node in the fringe such that n is on a
shortest path to an optimal goal G.

 f(G2) = g(G2) since h(G2) = 0

 g(G2) > g(G) since G2 is suboptimal
 f(G) = g(G) since h(G) = 0
 f(G2) > f(G) from above

 Optimality of A* (proof)

 f(G2) > f(G) from above

 h(n) ≤ h*(n) since h is admissible
 g(n) + h(n) ≤ g(n) + h*(n)
 f(n) ≤ f(G)
Hence f(G2) > f(n), and A* will never select G2 for expansion



 Consistent heuristics

 A heuristic is consistent or monotonous if

 for every node n, every successor n' of n generated by any action a,
satisfies h(n) ≤ c(n,a,n') + h(n')
 h(G) = 0 for each goal state G
 If h is consistent, we have
f(n') = g(n') + h(n')
= g(n) + c(n,a,n') + h(n')
≥ g(n) + h(n)
= f(n)
 i.e., f(n) is non-decreasing along any path.

 Theorem: Every consistent heuristic is also admissible

 Theorem: If h(n) is consistent, A* using GRAPH-

SEARCH is optimal
 Optimality of A*

 A* expands nodes in order of increasing f value

 Gradually adds "f-contours" of nodes
 Contour i has all nodes with f=fi, where fi < fi+1
 Properties of A*

Complete? Yes (unless there are infinitely many

nodes with f ≤ f(G) )
Time? Exponential
Space? Keeps all nodes in memory
Optimal? Yes
 Admissible heuristics

E.g., for the 8-puzzle:

 h1(n) = number of misplaced tiles
 h2(n) = total Manhattan distance
(i.e., no. of squares from desired location of each tile)

h1(S) = ?
h2(S) = ?

 Admissible heuristics

E.g., for the 8-puzzle:

 h1(n) = number of misplaced tiles
 h2(n) = total Manhattan distance
(i.e., no. of squares from desired location of each tile)

h1(S) = ? 8
h2(S) = ? 3+1+2+2+2+3+3+2 = 18

 Dominance

 If h2(n) ≥ h1(n) for all n (both admissible)

then h2 dominates h1
 h2 is better for search

 Typical search costs (average number of nodes expanded):

 d=12 IDS = 3,644,035 nodes

A*(h1) = 227 nodes
A*(h2) = 73 nodes
 d=24 IDS = too many nodes
A*(h1) = 39,135 nodes
A*(h2) = 1,641 nodes
 The A* procedure
2 8 3
m(A) d(x)
1 6 4 =4 =0
7 5

2 8 3
m(B)
2 8 3
m(C)
2 8 3
m(D) d(x)
The 8-puzzle with
1 6 4 =5 1 4 =3 1 6 4 =5
7 5 7 6 5 7 5
=1 optimistic heuristics.

2 8 3 2 3 2 8 3 A, ..., N: states;
m(E) m(F) m(G) d(x)
1 4 =3 1 8 4 =3 1 4 =4 =2 m = misplaced;
7 6 5 7 6 5 7 6 5
d = depth;
8 3 2 8 3 2 3 2 3 f(x) = m(x) + d(x).
m(H) m(I) m(J) m(K) d(x)
2 1 4 =3 7 1 4 =4 1 8 4 =2 1 8 4 =4 =3
7 6 5 6 5 7 6 5 7 6 5

1 2 3
m(L) d(x)
8 4 =1 =4
7 6 5

1 2 3 1 2 3
m(M) m(N) d(x)
8 4 =0 7 8 4 =2 =5
7 6 5 6 5
 :

Tutorial Question

Suppose you have the following search space

State next cost Heuristic Distance
A B 1
A C 4
A 37
B D 8
B E 3
B 42
C D 6 C 30
C F 2 D 19
D E 2 E 13
E G 4 F 45
F G 7
G 0
 Tutorial

 Draw the state space of this problem.

 Assume that the initial state is A and the goal state is G. Show
how each of the following search strategies would create a
search tree to find a path from the initial state to the goal state:
 Breadth-first search
 Depth-first search
 Uniform cost search
 Greedy search
 A* search
 At each step of the search algorithm, show which node is being
expanded, and the content of fringe. Also report the eventual
solution found by each algorithm.
Thank You