Artificial Intelligence

Lecture 5:
Uniformed Search

Tauseef Iftikhar

Punjab University College of Information Technology

Lahore.
Today’s Agenda

Recap

Uninformed Search
Recap

I Goal-base agents
I Problem Solving
I Goal formulation
I Problem formulation
I Problem formulation
I Initial state
I Actions
I Transition model
I Goal test
I Path cost
I State space vs Search space
I Search space regions
I Explored
I Frontier
I Unexplored
Recap

I Tree Search
I Graph Search
I Search Strategies Evaluation criteria
I Completeness
I Time Complexity
I Space Complexity
I Optimality
Time and space complexity are measured in terms of
I b: maximum branching factor of the search tree (actions per
state)
I d: depth of the solution
I m: maximum depth of the state space (may be ∞)
Two kind of search: uninformed and informed
Uninformed Search

I Also known as blind search.

I No additional information about states beyond that provided
in the problem definition.
I Just capable to distinguish goal state from nongoal state.
I 5 Search strategies come under the uninformed search
I Breadth-first search
I Uniform cost search
I Depth-first search
I Depth-limited search
I Iterative deepening depth-first search
Breadth-first search

All the nodes are expanded at a given depth in the search tree
before any nodes at the next level are expanded.
Breadth-first search
Breadth-first search
Breadth-first search
Breadth-first search
Evaluation of BFS
I Completeness:
Evaluation of BFS
I Completeness: Yes; if b is finite
Evaluation of BFS
I Completeness: Yes; if b is finite
I Time Complexity:
Evaluation of BFS
I Completeness: Yes; if b is finite
I Time Complexity: 1 + b + b 2 + b 3 + ... + b d = O(b d )
Evaluation of BFS
I Completeness: Yes; if b is finite
I Time Complexity: 1 + b + b 2 + b 3 + ... + b d = O(b d )
I Space Complexity:
Evaluation of BFS
I Completeness: Yes; if b is finite
I Time Complexity: 1 + b + b 2 + b 3 + ... + b d = O(b d )
I Space Complexity: O(b d )
Evaluation of BFS
I Completeness: Yes; if b is finite
I Time Complexity: 1 + b + b 2 + b 3 + ... + b d = O(b d )
I Space Complexity: O(b d )
I Optimality:
Evaluation of BFS
I Completeness: Yes; if b is finite
I Time Complexity: 1 + b + b 2 + b 3 + ... + b d = O(b d )
I Space Complexity: O(b d )
I Optimality:Yes; if the path cost is a nondecreasing function
of the depth of the node.
Evaluation of BFS
I Completeness: Yes; if b is finite
I Time Complexity: 1 + b + b 2 + b 3 + ... + b d = O(b d )
I Space Complexity: O(b d )
I Optimality:Yes; if the path cost is a nondecreasing function
of the depth of the node.
How bad is BFS?
Evaluation of BFS
I Completeness: Yes; if b is finite
I Time Complexity: 1 + b + b 2 + b 3 + ... + b d = O(b d )
I Space Complexity: O(b d )
I Optimality:Yes; if the path cost is a nondecreasing function
of the depth of the node.
How bad is BFS?

Time and Memory requirements for breadth-first search for a

branching factor b=10; 1 million nodes per second; 1,000 bytes per
Depth-first Search
DFS always expand the deepest node in the frontier of the search
tree.
Depth-first Search
DFS always expand the deepest node in the frontier of the search
tree.
Depth-first Search
DFS always expand the deepest node in the frontier of the search
tree.
Depth-first Search
DFS always expand the deepest node in the frontier of the search
tree.
DFS Search
DFS Evaluation
I Completeness:
DFS Evaluation
I Completeness: Yes; if d is finite. No in infinite search space.
DFS Evaluation
I Completeness: Yes; if d is finite. No in infinite search space.
I Time Complexity:
DFS Evaluation
I Completeness: Yes; if d is finite. No in infinite search space.
I Time Complexity: 1 + b + b 2 + b 3 + ... + b m = O(b m )
DFS Evaluation
I Completeness: Yes; if d is finite. No in infinite search space.
I Time Complexity: 1 + b + b 2 + b 3 + ... + b m = O(b m )
I Space Complexity:
DFS Evaluation
I Completeness: Yes; if d is finite. No in infinite search space.
I Time Complexity: 1 + b + b 2 + b 3 + ... + b m = O(b m )
I Space Complexity: O(bm) Linear complexity (needs to store
only a single path from the root to a leaf node, along with the
remaining unexpanded sibling nodes for each node on the
path. )
DFS Evaluation
I Completeness: Yes; if d is finite. No in infinite search space.
I Time Complexity: 1 + b + b 2 + b 3 + ... + b m = O(b m )
I Space Complexity: O(bm) Linear complexity (needs to store
only a single path from the root to a leaf node, along with the
remaining unexpanded sibling nodes for each node on the
path. )
I Optimality:
DFS Evaluation
I Completeness: Yes; if d is finite. No in infinite search space.
I Time Complexity: 1 + b + b 2 + b 3 + ... + b m = O(b m )
I Space Complexity: O(bm) Linear complexity (needs to store
only a single path from the root to a leaf node, along with the
remaining unexpanded sibling nodes for each node on the
path. )
I Optimality:No.
DFS Evaluation
I Completeness: Yes; if d is finite. No in infinite search space.
I Time Complexity: 1 + b + b 2 + b 3 + ... + b m = O(b m )
I Space Complexity: O(bm) Linear complexity (needs to store
only a single path from the root to a leaf node, along with the
remaining unexpanded sibling nodes for each node on the
path. )
I Optimality:No.
How bad is DFS?
DFS Evaluation
I Completeness: Yes; if d is finite. No in infinite search space.
I Time Complexity: 1 + b + b 2 + b 3 + ... + b m = O(b m )
I Space Complexity: O(bm) Linear complexity (needs to store
only a single path from the root to a leaf node, along with the
remaining unexpanded sibling nodes for each node on the
path. )
I Optimality:No.
How bad is DFS?

At depth 12=? , At depth 16=?

Depth-limited search

I To cope the problem of unbounded tree in depth first search,

predetermined depth limit l is used.
I However, it introduces additional source of incompleteness if
l <d
I It can be nonoptimal if l > d.
I Time complexity is O(b l ).
I Space complexity is O(bl).
Iterative deepening depth-first search

I Combines the benefit of DFS and BFS

I It find the best depth limit.
I idea: Iteratively increase the search limit until the depth of
the shallowest solution d is reached.
I Because most of the nodes are on the bottom of the search
tree, it not a big waste to iteratively re-generate the top
Iterative deepening
iterative deepening
iterative deepening
iterative deepening
Uniform cost search

I UCS expand the node n with the lowest path cost.

I It does not care about the number of steps a path has.
Uniform cost search
UCS search
UCS Evaluation

I Completeness:
UCS Evaluation

I Completeness: Yes; if cost of every step is greater than or

equal to some small positive constant .
UCS Evaluation

I Completeness: Yes; if cost of every step is greater than or

equal to some small positive constant .
I Time Complexity:
UCS Evaluation

I Completeness: Yes; if cost of every step is greater than or

equal to some small positive constant .
I Time Complexity:
I Let C ∗ be the cost of optimal solution.
I The effective path is roughly C ∗ /
∗
I O(b C / )
UCS Evaluation

I Completeness: Yes; if cost of every step is greater than or

equal to some small positive constant .
I Time Complexity:
I Let C ∗ be the cost of optimal solution.
I The effective path is roughly C ∗ /
∗
I O(b C / )
I Space Complexity:
UCS Evaluation

I Completeness: Yes; if cost of every step is greater than or

equal to some small positive constant .
I Time Complexity:
I Let C ∗ be the cost of optimal solution.
I The effective path is roughly C ∗ /
∗
I O(b C / )
∗ /
I Space Complexity: O(b C )
UCS Evaluation

I Completeness: Yes; if cost of every step is greater than or

equal to some small positive constant .
I Time Complexity:
I Let C ∗ be the cost of optimal solution.
I The effective path is roughly C ∗ /
∗
I O(b C / )
∗ /
I Space Complexity: O(b C )
I Optimality:
UCS Evaluation

I Completeness: Yes; if cost of every step is greater than or

equal to some small positive constant .
I Time Complexity:
I Let C ∗ be the cost of optimal solution.
I The effective path is roughly C ∗ /
∗
I O(b C / )
∗
I Space Complexity: O(b C / )
I Optimality:Yes, with same condition as for completeness
Exercise

What is the order of the visit of the nodes and the path returned
by BFS, DFS and UCS