Unit3:Informed and Uninformed search

Unit 3: Informed and Uninformed Search LH 8
Presented By : Tekendra Nath Yogi
Tekendranath@gmail.com
College Of Applied Business And Technology

Contd…
• Unit 3: Informed and Uninformed Search LH 8
– 3.1 Why search in AI?
– 3.2 Blind search (Un-informed search)
• 3.2.1 Breadth first search (BFS)
– Variations: Uniform cost search
• 3.2.2 Depth first search (DFS)
– Variations: Depth limited search, Iterative deepening DFS
– 3.3 Heuristic search (Informed search)
• 3.3.1 Hill climbing
– The Foothills Problem
– The Plateau Problem
– The Ridge Problem
• 3.3.2 Greedy (Best-first) search
• 3.3.3 A* algorithm search
• 3.3.4 Means-Ends Analysis: Household ROBOT, Monkey Banana Problem
212/25/2018 By: Tekendra Nath Yogi

Contd…
• Unit 3: Informed and Uninformed Search contd…
– 3.4 General Problem Solving (GPS): Problem solving agents
• 3.4.1 Constraint satisfaction problem
– Constraint Satisfaction Search
– AND/OR trees
– The bidirectional search
– Crypto-arithmatics
– 3.5 Game playing and AI
• 3.2.1 Game Trees and Mini-max Evaluation
• 3.2.2 Heuristic Evaluation
• 3.2.3 Min-max algorithm (search)
• 3.2.4 Min-max with alpha-beta
• 3.2.5 Games of chance
• 3.2.6 Game theory

Four General steps in problem solving
• Problem solving is a systematic search through a range of possible actions in
order to reach some predefined goal or solution.
• For problem solving a kind of goal based agent called problem solving agents
are used.
• This agent first formulates a goal and a problem, searches for a sequence of
actions that would solve the problem, and then executes the actions one at a
time. When this is complete, it formulates another goal and starts over.
• This over all process is described in the following four steps:
– Goal formulation
– Problem formulation
– Search
– Execute

Contd…
• Goal formulation:
– Intelligent agent maximize their performance measure by adapting a goal
and aim at satisfying it.
– Goal help organize behavior by limiting the objectives that the agent is
trying to achieve and hence the actions it needs to consider.
– Goal are the set of world states in which the goal is satisfied.
– Therefore, during goal formulation step, specify what are the successful
world states.

Contd…
• Problem Formulation:
– Problem formulation is the process of deciding what actions and states to
consider, given a goal.
– Therefore, the agent‟s task is to find out how to act, now and in the future,
so that it reaches a goal state.
– Before it can do this, it needs to decide(or we need to decide on its
behalf) what sorts of actions and states it should consider.

Contd…
• Search a solution:
– The process of looking for a sequence of actions that reaches the goal is
called a searching.
– search algorithm takes a problem as a input and returns a solution in the
form of an action sequence.

Contd…
• Execution:
– once a solution is found, the actions it recommends can be carried out.
This is called the execution phase.
– once a solution has been executed , the agent will formulate a new goal.

Problem Formulation
• A problem can be defined formally by five components:
– Initial state
– Actions
– Transition model
– Goal Test
– Path Cost

Contd…
• Initial state: The state from which agent start.
• Actions: A description of the possible actions available to the agent. During
problem formulation we should specify the all possible actions available for
each state „s‟.
• Transition model: A description of what each action does is called the
transition model. For formulating transition model in problem formulation we
take state „s‟ and action „a‟ for that state and then specify the resulting state
„s‟
• Goal Test: Determine whether the given state is goal state or not.
• Path Cost: Sum of cost of each path from initial state to the given state

One way to formally define a problem: State space Representation
• The set of all states reachable from the initial state by any sequence of actions is
called state space.
• The state space forms a directed graph in which nodes are states and the links
between nodes are actions.
• A State space representation allows for the formal definition of a problem which
makes the movement from initial state to goal state quite easy.
• Disadvantage: it is not possible to visualize all states for a given problem. Also,
the resources of the computer system are limited to handle huge state space
representation.

Contd…
• State Space representation of Vacuum World Problem: Vacuum world
can be formulated as a problem as follows:
– States: The state is determined by both the agent location and the dirt locations. The
agent is in one of two locations, each of which might or might not contain dirt.
– Initial state: Any state can be designated as the initial state.
– Actions: In this simple environment, each state has just three actions: Left, Right,
and Suck. Larger environment may might also include Up and Down.
– Transition Model: The actions have their expected effects , except that moving
Left in leftmost square, moving Right in the rightmost square, and sucking in a
clean square having no effect. The complete state space is shown in figure below.
– Goal Test: This checks whether all the squares are clean.
– Path cost: Each step costs 1, so the path cost is the number of steps in the path.

Contd…
Fig: State space representation for the vacuum world
Here, links denote actions: L: left, R= Right, S= Suck

Searching For Solution
• Having formulated some problems, we now need to solve them.
• To solve a problem we should perform a systematic search through a range of
possible actions in order to reach some predefined goal or solution.
• A solution is an action sequence, so search algorithms works by considering
various possible action sequences.
• The possible action sequences starting at the initial state form a search tree
with the initial state at the root; the branches are actions and the nodes
correspond to states in the state space of the problem.

Contd…
• General search:
– The searching process starts from the initial state (root node )and
proceeds by performing the following steps:
• Check whether the current state is the goal state or not?
• Expand the current state to generate the new sets of states.
• Choose one of the new states generated for search depending upon search
strategy.
• Repeat step 1 to 3 until the goal state is reached or there are no more state to
be expanded.

Contd…
• The Importance of Search in AI:
– Many of the tasks underlying AI can be phrased in terms of a search for
the solution to the problem at hand.
– Many goal based agents are essentially problem solving agents which
must decide what to do by searching for a sequence of actions that lead to
their solutions.
– For the production systems, need to search for a sequence of rule
applications that lead to the required fact or action.
– For neural network systems, need to search for the set of connection
weights that will result in the required input to output mapping.

Contd…
• Measuring problem Solving Performance:
– The performance of the search algorithms can be evaluated in four ways:
– Completeness: An algorithm is said to be complete if it definitely finds
solution to the problem, if exist.
– Time complexity: How long does it take to find a solution? Usually
measured in terms of the number of nodes expanded during the search.
– Space Complexity: How much space is used by the algorithm? Usually
measured in terms of the maximum number of nodes in memory at a time
– Optimality/Admissibility: If a solution is found, is it guaranteed to be an
optimal one? For example, is it the one with minimum cost?

Classes of Search methods
• There are two broad classes of search methods:
• Uninformed (or blind) search methods:
– Strategies have no additional information about states beyond that provided in
the problem definition. All they can do is generate successors and distinguish a
goal state from a non-goal state.
– All search strategies are distinguished by the order in which nodes are expanded .
• Heuristically informed search methods.
– Strategies that know whether one non-goal state is “more promising” than another
are called informed or heuristic search strategies.
– I.e., In the case of the heuristically informed search methods, one uses domain-
dependent (heuristic) information in order to search the space more efficiently.

Uninformed (Or Blind) Search Methods:
• Blind search do not have additional information about state beyond the
problem definition to search a solution in the search space.
• It proceeds systematically by exploring nodes either randomly or in
some predetermined order.
• Based on the order in which nodes are expanded, it is of the following
types:
– Breadth First search (BFS)
• Variation: Uniform cost search
– Depth First search (DFS)
• Variations: Depth limit search, Iterative deepening DFS.
– Bidirectional search
19Presented By: Tekendra Nath YogiIT 228 - Intro to AI

Breadth First Search
• Breadth First search is a simple strategy in which the root node is expanded
first, then all the successors of the root node are expanded next, then their
successors, and so on.
• In general, All nodes are expended at a given depth in the search tree before
any nodes at the next level are expanded until the goal reached.
– I.e., Expand shallowest unexpended node.
• The search tree generated by the BFS is shown in figure below:
Fig: search Tree For BFS
Note: We are using the convention that the alternatives are tried in the left to right
order.

Breadth-first search

Uniform Cost Search
• The search begins at root node. The search continues by visiting the next node
which has the least total cost from the root node. Nodes are visited in this
manner until a goal is reached.
• Now goal node has been generated, but uniform cost search keeps going,
choosing a node (with less total cost from the root node to that node than the
previously obtained goal path cost) for expansion and adding a second path.
• Now the algorithm checks to see if this new path is better than the old one; if it
is so the old one is discarded and new one is selected for expansion and the
solution is returned.

Uniform cost search example1 (Find path from A to E)
• Expand A to B,C,D
• The path to B is the cheapest one with path cost 2.
• Expand B to E
– Total path cost = 2+9 =11
• This might not be the optimal solution since the path
AC as path cost 4 ( less than 11)
• Expand C to E
– Total path cost = 4+5 =9
• Path cost from A to D is 10 ( greater than path cost, 9)
Hence optimal path is ACE

• The graph below shows the step-costs
for different paths going from the
start (S) to the goal (G).
• Use uniform cost search to find the
optimal path to the goal.
Presented By: Tekendra Nath Yogi 26
Home work: Uniform cost search
IT 228 - Intro to AI

Depth First Search
• DFS also begins by expanding the initial node.
• Looks for the goal node among all the children of the current node before using
the sibling of this node
• i.e. expand deepest unexpanded node(expand most recently generated deepest
node first.).
• The search tree generated by the DFS is shown in figure below:

Depth-first search example
• Expand deepest unexpanded node
• Here initial state is A and goal state is M

• But this type of search can go on and on, deeper and deeper into tree
search space and thus, we can get lost. This is referred to as blind alley.

Depth-limited search(Same as DFS if L= ∞)
• Depth limit search is depth-first search with depth limit L.
– i.e., nodes at depth L are treated as they have no successors.
• The depth limit solves the infinite path problem of DFS by placing limit on the
depth.
• Yet it introduces another source of problem if we are unable to find good guess
of L. Let d is the depth of shallowest solution.
– If L < d then incompleteness results because no solution within the depth limit.
– If L > d then not optimal.

Depth-limited search Example 1
• Depth limited search= DFS+ limit for the depth
– Let goal node= 11 and limit= 2
– Trace the path to the goal node using Depth limited search .

Depth-limited search Example 1
• Depth limited search= DFS+ limit for the depth
• No path is found from root node to goal node because L< d.

Iterative deepening search
• It is a general strategy to find best depth limit L.
• It begin by performing DFS to a depth of zero, then depth of one, depth of
two , and so on until a solution is found or some maximum depth is reached.
• It is similar to BFS in that it explores a complete layer of new nodes at each
iteration before going to next layer.
• It is similar to DFS for a single iteration.
• It is preferred when there is a large search space and the depth of a solution is
not known.
• But it performs the wasted computation before reaching the goal depth.

Iterative Deepening search example
• Here initial state is A and goal state is M.
• Let we don‟t know the depth of M then the Iterative deepening search proceeds
as follows:

Iterative deepening search l =0

Bidirectional search
• This search is used when a problem has a single goal state that is given
explicitly and all the node generation operators have inverses.
• So it is used to find shortest path from an initial node to goal node
instead of goal itself along with path.
• It works by searching forward from the initial node and backward from
the goal node simultaneously, by hoping that two searches meet in the
middle.
• Check at each stage if the nodes of one have been generated by the
other,. i.e., they meet in the middle.
• If so, the path concatenation is the solution.
Presented By: Tekendra Nath Yogi 49IT 228 - Intro to AI

Bidirectional search contd..
• Advantages:
– Only slight modification of DFS and
BFS can be done to perform this
search.
– Theoretically effective than
unidirectional search.
• Disadvantage:
– Problem if there are many goal states.
– Practically inefficient due to additional
overhead to perform intersection
operation at each point of search
Presented By: Tekendra Nath Yogi 50IT 228 - Intro to AI

Drawbacks of uninformed search :
• Criterion to choose next node to expand depends only on a global criterion:
level.
• Does not exploit the structure of the problem.

Heuristic Search:
• Heuristic Search Uses domain-dependent (heuristic) information beyond the
definition of the problem itself in order to search the space more efficiently.
• Ways of using heuristic information:
– Deciding which node to expand next, instead of doing the expansion in a strictly
breadth-first or depth-first order;
– In the course of expanding a node, deciding which successor or successors to
generate, instead of blindly generating all possible successors at one time;
– Deciding that certain nodes should be discarded, or pruned, from the search
space.

Contd…
• Informed Search Define a heuristic function, h(n), that estimates the
"goodness" of a node n.
• The heuristic function is an estimate, based on domain-specific information
that is computable from the current state description, of how close we are to a
goal.
• Specifically, h(n) = estimated cost (or distance) of minimal cost path from
state „n‟ to a goal state.

Contd…
• A) Best-First Search:
– Best first search uses an evaluation function f(n) that gives an indication
of which node to expand next for each node.
– A key component of f(n) is a heuristic function, h(n),which is a additional
knowledge of the problem.
– Based on the evaluation function best first search can be categorized into
the following categories:
• Greedy best-first search
• A*search

Contd…
• Greedy Best First Search :
– Greedy best first search expands the node that seems to be closest to the
goal node.
– Evaluation function based on Heuristic function is used to estimate
which node is closest to the goal node.
– Therefore, Evaluation function f(n) =heuristic function h(n)= estimated
cost of the path from node n to the goal node.
– E.g., hSLD(n) = straight-line distance from n to goal
– Note: g(root)= 0 and h(goal) = 0

Contd…
• Example1 To illustrate Greedy Best- First Search:
– For example consider the following graph
– Straight Line distances to node G (goal node) from other nodes is given below:
– Let H(n)= Straight Line distance
– Now Greedy Search operation is done as below:

Contd…
• Start at node „s‟, the start state
• Children of s= {B(4), D(5)}
• Therefore, best = B

Contd…
• Children of B= {E(7), C(3)}
• Considered= {D(5), E(7), C(3)}
• Therefore, Best= C

Contd…
• Children of C= {D(5), G(0)}
• Considered= {D(5), E(7), G(0)}
• Therefore, Best= G, is the goal node.

Contd…
• A * Search :
– A* is a best first , informed search algorithm. The search begins at root node.
The search continues by visiting the next node which has the least
evaluation.
– It evaluates nodes by using the following evaluation function
• f(n) = h(n) + g(n) = estimated cost of the cheapest solution through n.
• Where, g(n): the actual shortest distance traveled from initial node to current
node , It helps to avoid expanding paths that are already expensive
• h(n): the estimated (or "heuristic") distance from current node to goal, it estimate
which node is closest to the goal node.
– Nodes are visited in this manner until a goal is reached.

Contd…
• Example to illustrate A* Search:
– For example consider the following graph
– Straight Line distances to node G (goal node) from other nodes is given
below:
– Labels in the graph shows actual distance.
– Let H(n)= Straight Line distance
– Now A* Search operation is done as below

Contd…
• A* algorithm starts at s, the start state.
• Children of S={B, D}
• Now, evaluation function for each child of S is:
– f(B) =g(B) + h(B) =9+4=13
– f(D)=g(D)+h(D)= 8+6=14
• Here, candidate node for expansion are{B, D}, among these candidate nodes
the node B is least evaluated, so it is selected for expansion.

Contd…
• Child of B ={E, C}
• Now, evaluation for each child of B are
– F(E)= (9+8)+7=24
– F(c)=(9+7)+3=19
• Here, candidate nodes for expansion are{E,C, and D}
• Among these candidate the node D is least evaluated, so it is selected for expansion
63Presented By: Tekendra Nath Yogi
f(n)= 14H(n)= 4

Contd….
• Child of D={G, C}
Evaluation for the child of D are
• Here, the candidate nodes for expansion are{ E, C,G and C}.
• Among these candidate the node C child of D is least
evaluated, so It is selected for expansion.
H(n)= 6

Contd…
• Child of C= {G,B}
Evaluation for the child of C are
• Here the candidate for expansion are{E,C,G,G, and B}.
• Among these candidate the node G, which is child of D is least evaluated, so
it is selected for expansion, but it is the goal node, hence we are done
H(n)= 6
H(n)= 3

Contd…
H(n)= 6
H(n)= 3

Contd…
• B) Local search algorithms:
– Local search algorithms operate using a single current node and move
only to neighbors of this node rather than systematically exploring paths
from an initial state. E.g. Hill climbing.
– These algorithms are suitable for problems in which all that matter is the
solution state, not the path cost to reach it. Typically, the paths followed
by the search are not retained.
– Although local search algorithms are not systematic, they have two key
advantages:
• They use very little memory.
• They can often find reasonable solutions in large or infinite state spaces for
which systematic algorithms are unsuitable.

Contd…
• Hill Climbing Search:
– Hill climbing can be used to solve problems that have many solutions,
some of which are better than others.
– It starts with a random (potentially poor) solution, and iteratively makes
small changes to the solution, each time improving it a little. When the
algorithm cannot see any improvement anymore, it terminates.
– Ideally, at that point the current solution is close to optimal, but it is not
guaranteed that hill climbing will ever come close to the optimal solution.
– Note: The algorithm does not maintain a search tree and does not look
ahead beyond the immediate neighbors of the current state.

Contd…

Contd…
• Hill climbing suffers from the following problems:
– The Foot- hills problem(local maximum)
– The Plateau problem
– The Ridge problem

Contd…
• Foot- Hill problem:
– Local maximum is a state which is better than all of its neighbors but is
not better than some other states which are farther away.
– At local maxima, all moves appear to make the things worse.
– This problem is called the foot hill problem.
– Solution: Backtrack to some earlier node and try going to different
direction.

Contd…
• The Plateau problem:
– Plateau is a flat area of the search space in which a whole set of
neighboring states have the same value.
– On plateau, it is not possible to determine the best direction in which to
move by making local comparison.
– Such a problem is called plateau problem.
– Solution: Make a big jump in some direction to try to get a new section
of the search space .

Contd…
• The Ridge problem:
– Ridge is an area of the search space which is higher than the surrounding areas
and that itself has a slope.
– Due to the steep slopes the search direction is not towards the top but towards the
side(oscillates from side to side).
– Such a problem is called Ridge problem.
• Solution: Apply two or more rules such as bi-direction search before doing
the test.

Mean-End Analysis
• Mean end analysis is a problem solving technique. Allows us to solve the major
parts of a problem first and then go back and solve the smaller problems that
arise while assembling the final solution.
• Technique:
– The Mean End analysis process first detects the differences between current
state and the goal state.
– Once such a difference is isolated, an operator that can reduce the difference
has to be found.
– But sometimes it is not possible to apply this operator to the current state.
So, we try to get a sub- problem out of it and try to apply our operator to this
new state.
– If this also does not produce the desired goal state then we try to get second
sub-problem and apply this operator again. This process may be continued.

Contd…
• Problem:
• Solution:

Contd…
• Why it is called mean end analysis?
– Given a description of the desired state of the world(the end ).
– It works by selecting and using operators that will achieve it(the means).
– Hence the name, mean end analysis(MEA).

Contd…
• Mean End Analysis( Household Robot):
– Consider a simple household robot domain. Problem Given to the robot
is: Move a desk with two things on it from one room to another. The
objects on top must also be moved.

Contd…
• The available operators are shown below along with their pre-conditions and results.
• Table below shows when each of the operators is appropriate:

Contd…
The robot solves this problem as follows:
• The main difference between the initial state and the goal state would be the
location of the desk.
• To reduce this difference, either PUSH or CARRY could be chosen.
• If CARRY is chosen first, its preconditions must be met. This results in two
more differences that must be reduced:
– The location of the robot and the size of the desk.
– The location of the robot is handled by applying WALK, but there are no
operators that can change the size of an object, So this path leads to a dead-end.

Contd…
• Following the other branch, we attempt to apply PUSH. Figure below shows
the robot‟s progress at this point.
Fig : The progress of mean end analysis method.
• Now the differences between A and B and Between C and D must be reduced.

Contd…
• PUSH has preconditions:
– the robot must be at the desk, and
– the desk must be clear.
• The robot can be brought to the correct location by using WALK .
• And the surface of the desk can be cleared by two uses of the PICKUP,.
• But after one PICKUP, An attempt to do the second results in another
difference(the arm must be empty).
• PUTDOWN can be used to reduce the difference.

Contd…
• Once PUSH is performed, the problem state is closed to the goal state, but not
quite. The object must be placed back on the desk. PLACE will put them
there. The progress of the robot at this point is as shown in figure below:
Fig: More progress on the mean end analysis method.
• The final difference between C and E can be reduced by using WALK to get
the robot back to the object followed by PICKUP and CARRY.

Contd…
• Mean-End Analysis:(Monkey Banana Problem):
– The monkey is in a closed room in which there is a small Box.
– There is a bunch of bananas hanging from the ceiling but the monkey cannot reach
them.
– Assuming that the monkey can move the Box and if the monkey stands on the Box
the monkey can reach the bananas.
– Establish a method to instruct the monkey on how to capture the bananas.

Contd…
• Representing the World:
– In the Monkey Banana problem we have:
• objects: a monkey, a box, the bananas, and a floor.
• locations: we‟ll call them a, b, and c.
• relations of objects to locations. For example:
– the monkey is at location a;
– the monkey is on the floor;
– the bananas are hanging;
– the box is in the same location as the bananas.
– Formally Representing the relations of objects to locations as:
– at(monkey, a).
– on(monkey, floor).
– status(bananas, hanging).
– at(box, X), at(bananas, X).

Contd…
• Initial State:
at(monkey, a),
on(monkey, floor),
at(box, b),
on(box, floor),
at(bananas, c),
status(bananas, hanging).
• Goal State:
on(monkey, box),
on(box, floor),
at(monkey, c),
at(box, c),
at(bananas, c),
status(bananas, grabbed).

Contd…
• All Operators:
Operator Preconditions Results
go(X,Y) at(monkey,X) at(monkey, Y)
on(monkey, floor)
push(B,X,Y) at(monkey,X) at(monkey, Y)
at(B,X) at(B,Y)
on(monkey, floor)
on(B,floor)
climb_on(B) at(monkey,X) on(monkey, B)
at(B,X)
on(monkey,floor)
on(B,floor)
grab(B) on(monkey,box) status(B, grabbed)
at(box,X)
at(B,X)
status(B, hanging)

Contd…
• Instructions To the Monkey to Grab the banana:
– First of all Monkey should move form location a to location b.
• Go(a, b)
– Then push the Box form location b to location a.
• Push(B, b, a)
– Push the Box again from location a to location c.
• Push(B, a, c)
– Monkey should climb on the Box
• Climb_on(B)
– Then Grab the banana.
• Grab(B)

Constraint satisfaction problems
• A CSP consists of three components: X, D and C
– X is a Finite set of variables {X1, X2, …, Xn }
– D is a set of Nonempty domain of possible values for each variable { D1,
D2, … Dn }
– C is a Finite set of constraints {C1, C2, …, Cm }
– Each constraint Ci limits the values that variables can take, e.g., X1 ≠ X2

Contd…
• In CSP:
– State: A state is defined as an assignment of values to some or all
variables.
– Consistent assignment: consistent assignment is an assignment that does
not violate the constraints.
– An assignment is complete when every variable is assigned a value.
– A solution to a CSP is a complete assignment that satisfies all constraints
i.e., complete and consistent assignment.

Contd…
• Example (Map-Coloring problem): Given, a map of Australia showing
each of states and territories. The task is color each region either red, green,
or blue in such a way that no neighboring regions have the same color.
• This problem can be formulated as CSP as
follows:
– Variables: WA, NT, Q, NSW, V, SA, T
– Domains: Di = { red, green, blue}
– Constraints: adjacent regions must have different
colors
• e.g., WA ≠ NT

Contd…
• Solutions are complete and consistent assignments,
– e.g., WA = red, NT = green, Q = red, NSW = green, V = red, SA = blue,
T = green

Contd…
• CSP as a standard search problem:
– A CSP can easily expressed as a standard search problem, as
follows:
• Initial State: the empty assignment {}, in which all variables are
unassigned..
• Successor function: Assign value to unassigned variable provided that
there is not conflict.
• Goal test: the current assignment is complete and consistent.
• Path cost: a constant cost for every step.

Contd…
• Constraint satisfaction search:
– For searching a solution of CSPs following algorithms can be used:
• Backtracking search: Works on partial assignments.
• Local search for CSPs: works on complete assignment.

Contd…
• Backtracking search:
– The term backtracking search is used for a depth first search that chooses
values for one variable at a time and backtracks when a variable has no
legal values left to assign.
– The algorithm repeatedly choose an unassigned variable, and then tries all
values in the domain of that variable in turn, trying to find a solution.
– If an inconsistency is detected, then backtrack returns failure, causing the
previous call to try to another values.

95
Backtracking example
Presented By: Tekendra Nath YogiIT 228 - Intro to AI

96

97

98

Contd…
• Local search for CSPs:
– They use a complete state formulation:
• The initial state assigns a value to every variable, and the search
changes the value of one variable at a time because the initial guess
violates several constraints.
– E.g.:
• First randomly select any conflicted variable.
• Then choose the value of that conflicted variable that violates the
fewest constraints.

Crypt-arithmetic Problem
• Many problems in AI can be considered as problems of constraint satisfaction,
in which the goal state satisfies a given set of constraint.
 Example of such a problem is Crypt-Arithmetic problem (a mathematical
puzzle ), in which the goal state(solution) satisfies the following constraints:
– Values are to be assigned to letters from 0 to 9 only.
– No two letters should have the same value.
– If the same letter occurs more than once, it must be assigned the same digit each
time.
– The sum of the digits must be arithmetically correct with the added restriction that
no leading zeroes are allowed.

Contd…
• Example1: Solve the following puzzle by assigning numeral (0-9) in such a
way that each letter is assigned unique digit which satisfy the following
addition.

Contd…
• Solution: Here,
• Variables: {F, T, U, W, R, O, c1, c2, C3 }
• Domains: {0,1,2,3,4,5,6,7,8,9}
• Constraints: Alldiff (F,T,U,W,R,O)
• where c1, c2, and c3 are auxiliary variables representing the digit (0 or 1) carried over
into the next column.
c3 c2 c1

Contd…
• Here we are adding two three letters words but getting a four letters word. This indicates
that F= c3=1
• Now, c2+T+T= O+ 10……. Because c3=1
• C2 can be 0 or 1 . Let c2=0 then T should be > 5 i.e T can be {6, 7,8,9}
• Let T= 9 then C2+T+T= O+10 0+9+9=O+10 from this O= 8
• Now O+O=R+10 8+8= R+10 From this R=6
• Now, c1+W+W=U here c1=1 and U and W can be {2,3,4,5,7}
• But , c2= 0 so let W= 2 then 1+2+2=U i.e., U=5
• Now replacing each letter in the puzzle by its corresponding digit and testing their
arithmetic correctness:
• This assignment satisfies all the constraint so this is the final solution

Contd…
• Example2: Solve the following puzzle by assigning numeral (0-
9) in such a way that each letter is assigned unique digit which
satisfy the following addition.
• Solution: Here,
• Variables: {F,O,U,R,E,I,G,H,T, c1, c2, C3 ,c4 }
• Domains: {0,1,2,3,4,5,6,7,8,9}
• Constraints: Alldiff (F,O,U,R,E,I,G,H,T)
• where c1, c2, and c3 are auxiliary variables representing the digit (0 or 1) carried over
into the next column.

Contd…
• Here we are adding two three letters words but getting a four letters word. This
indicates that E= c4=1 ….. Because E is left most letter so it should not be 0.
• Now c3+F+F= I+10 , here c3 can be 0 or 1 and F should be greater than 5. i.e
{6,7,8,9}
• Let c3= 0 and F= 9 then 0+9+9=I+10 from this I=8
• Now, c2+O+O=G…… since c3=0
• C2 can be 0 or 1 and O can be {2,3,4}. Let c2= 0 and O =2 Then G= 4.
• R+R= T here R can be {3,5,6,7}
• If we let R= 3 this leads to dead end so let R=5 then T=0 and c1=1
• C1+U+U=H here c1= 1 and c2=0 and U can be{3}
• Form this U=3 then H=7

Contd…
• Initial problem state:
– S = ? M= ? C1 = ?
– E = ? O = ? C2 = ?
– N = ? R = ? C3 = ?
– D = ? E = ? C4 = ?
• Goal states: A goal state is a problem state in which all letters have been assigned a
digit in such a way that all constraints are satisfied

Contd…

Home Work
• Solve the following puzzles by assigning numeral (0-9) in such a way that
each letter is assigned unique digit which satisfy the following addition.
1.
2.
3.
4.

Problem Reduction: AND/ OR Tree
• The basic intuition behind the method of problem reduction is:
– Reduce a hard problem to a number of simple problems and, when each of
the simple problems is solved, then the hard problem has been solved.
• To represent problem reduction we can use an AND-OR tree. An AND–OR
tree is a graphical representation of the reduction of problems to conjunctions
and disjunctions of sub-problems.
• Example:
represents the search space for solving the
problem P, using the problem-reduction
methods:
P if Q and R
P if S
Q if T
Q if U

Contd…
• Example1
• Example2:

Contd….
• Searching AND/OR Tree:
• To find a solution in AND–OR tree, an algorithm similar to Best first search
algorithm is used but with the ability to handle AND arc appropriately.
• This algorithm evaluates the nodes based on the following evaluation function.
• If the node is OR node then cost function f(n)= h(n)
• If the node is AND node then cost function is the sum of costs in AND node.
– f(n) = f(n1) + f(n2) + …. + f(nk)= h(n1) + h(n2) + …. + h(nk)
– Where, n1, n2,…. Nk are AND nodes.

Contd…
• Here, the number show the value of the heuristic function at that node.
• In the figure the minimal is B which is the value of 3. But B forms a part of the AND
tree so we need to consider the other branch also of this AND tree i.e., C which has a
weight of 4. So, our estimate now is (3+4) = 7.
• Now this estimate is more costlier than that of branch D i.e., 5. So we explore node
D instead of B as it has the lowest value.
• This process continues until either a solution is found or all paths led to dead ends,
indicating that there is no solution.

Adversarial Search(Game Playing)
• Competitive environments in which the agents goals are in conflict, give rise
to adversarial search, often known as games.
• In AI, games means deterministic, fully observable environments in which
there are two agents whose actions must alternate and in which utility values
at the end of the game are always equal and opposite.
– E.g., If first player wins, the other player necessarily loses
• This Opposition between the agent‟s utility functions make the situation
adversarial.

Contd…
• Games as Adversarial Search:
– A Game can be formally defined as a kind of search problem with the
following elements:
• States: board configurations.
• Initial state: the board position and which player will move.
• Successor function: returns list of (move, state) pairs, each indicating
a legal move and the resulting state.
• Terminal test: determines when the game is over.
• Utility function: gives a numeric value in terminal states
– (e.g., -1, 0, +1 for loss, tie, win)

Contd…
• Game Trees: Problem spaces for typical games represented as trees, in which:
– Root node: Represents the state before any moves have been made.
– Nodes: Represents possible states of the games. Each level of the tree has
nodes that are all MAX or all MIN; nodes at level i are of the opposite kind
from those at level i+1. and
– Arcs: Represents the possible legal moves for a player. Moves are
represented on alternate levels of the game tree so that all edges leading
from root node to the first level represent moves for the first(MAX) player
and edges from the first level to second represents moves for the
second(MIN) player and so on.
– Terminal nodes represent end-game configurations.

Contd…
• Evaluation Function:
– An evaluation function is used to evaluate the "goodness" of a game
position.
– i.e., estimate of the expected utility of the game position
– The performance of a game playing program depends strongly on the
quality of its evaluation function.
– An inaccurate evaluation function will guide an agent toward positions
that turn out to be lost.

Contd…
• A good evaluation function should:
• Order the terminal states in the same way as the true utility function:
• i.e., States that are wins must evaluate better than draws, which in
turn must be better than losses. Otherwise, an agent using the
evaluation function might err even if it can see ahead all the way to
the end of the game.
• For non-terminal states, the evaluation function should be strongly
correlated with the actual chances of winning.
• The computation must not take too long i.e., evaluate faster.

Contd…
• An example (partial) game tree for Tic-Tac-Toe:
• f(n) = +1 if the position is a win for
X.
• f(n) = -1 if the position is a win for
O.
• f(n) = 0 if the position is a draw.

Contd…
• There are two players denoted by X and O. They are alternatively writing their letter in
one of the 9 cells of a 3 by 3 board. The winner is the one who succeeds in writing three
letters in line.
• The game begins with an empty board. It ends in a win for one player and a loss for the
other, or possibly in a draw.
• A complete tree is a representation of all the possible plays of the game. The root node is
the initial state, in which it is the first player's turn to move (the player X).
• The successors of the initial state are the states the player can reach in one move, their
successors are the states resulting from the other player's possible replies, and so on.
• Terminal states are those representing a win for X, loss for X, or a draw.
• Each path from the root node to a terminal node gives a different complete play of the
game. Figure given above shows the search space of Tic-Tac-Toe.

Mini-max search algorithm
• Mini-max search algorithm is a game search algorithm with the application of
DFS procedure.
• It assumes:
• Both the player play optimally from there to the end of the game.
• A suitable value of static evaluation(utility) is available on the leaf nodes.
• Given the value of the terminal nodes, the value of each node (Max and MIN) is
determined by (back up from) the values of its children until a value is
computed for the root node.
– For a MAX node, the backed up value is the maximum of the values associated
with its children
– For a MIN node, the backed up value is the minimum of the values associated with
its children

Contd…
• Mini-max search Example:

Contd…

Contd…
• Limitations of Mini-max search:
– Mini-max search traverse the entire search tree but it is not always
feasible to traverse entire tree.
– Time limitations

Alpha-beta pruning
• We can improve on the performance of the mini-max algorithm through
alpha-beta pruning.
• Basic idea: If a move is determined worse than another move already
examined, then there is no need for further examination of the node.
• For Example: Consider node n in the tree.
• If player has a better choice at:
– Parent node of n
– Or any choice point further up
• Then n is never reached in play.
• So, When that much is known about n, it can
be pruned.

Contd…
• Example:

Contd…
• Alpha-Beta pruning procedure:
– Traverse the search tree in depth-first order. During traversing Alpha and
Beta values inherited from the parent to child never from the children.
Initially alpha =-infinity and always try to increase, and beta=+infinity and
always try to decrease. Alpha value updates only at max node and beta value
update only at min node.
– Max player :
• Val> Alpha?(val is Value back up form the children of max player)
– Update Alpha
• Is Alpha>= Beta?
– Prune (called alpha cutoff)
• Return Alpha
– MIN player:
• Val< Beta? (val is Value back up form the children of min player)
– Update Beta
• Is Alpha>= Beta?
– Prune ( called beta cutoff)
• Return Beta

Contd…
• Alpha-Beta pruning example:

Contd…

Games of Chance
• In the game with uncertainty Players include a random element(roll dice, flip
a coin, etc. ) to determine what moves to make
• i.e., Dice are rolled at the beginning of a player‟s turn to determine the
legal moves. Such games are called game of chance.
• Chance games are good for exploring decision making in adversarial
problems involving skill and luck.

Contd…
• Game Trees with Chance Nodes:
• The game tree in the chance game include chance nodes in addition to
MAX and MIN nodes. Chance nodes (shown as circles) represent the dice
rolls.
•The branches leading from each chance node denote the possible dice rolls;
each branch is labeled with the roll and its probability.
Max
min
Chance
Terminal node
Chance

Contd…
• Algorithm for chance Games:
• Generalization of minimax for games with chance nodes. Also known as
Expectiminimax give perfect play
– If the state is terminal node then Return the utility value.
– if state is a MAX node then return highest Expectiminimax – Value of
Successors
– if state is a MIN node then return lowest Expectiminimax – Value of
Successors
– if state is a CHANCE node then
•For chance nodes „C‟ over a max node, compute: epectimax(C) =
Sumi(P(di) * maxvalue(i))
•For chance nodes „C‟ over a min node compute: epectimin(C) =
Sumi(P(di) * minvalue(i))

Contd…
• Example:

What is Game Theory?
• Game theory is a study of how to mathematically determine the best strategy
for given conditions in order to optimize the outcome
• Finding acceptable, if not optimal, strategies in conflict situations.
• Abstraction of real complex situation
• Game theory is highly mathematical
• Game theory assumes all human interactions can be understood and navigated
by presumptions.

Contd…
• Why is game theory important?
– All intelligent beings make decisions all the time.
– AI needs to perform these tasks as a result.
– Helps us to analyze situations more rationally and formulate an
acceptable alternative with respect to circumstance.
– Useful in modeling strategic decision-making
• Games against opponents
• Games against nature
– Provides structured insight into the value of information

Homework
• Explain the differences and similarities between depth-first search and
breadth-first search. Give examples of the kinds of problems where each
would be appropriate.
• Explain what is meant by the following terms in relation to search methods:
– complexity
– completeness
– Optimality
• Provide a definition of the word “heuristic.” In what ways can heuristics be
useful in search? Name three ways in which you use heuristics in your
everyday life.
• Explain the components of the path evaluation function f(node) used by
A*. Do you think it is the best evaluation function that could be used? To
what kinds of problems might it be best suited? And to what kinds of
problems would it be worst suited?
12/25/2018 By: Tekendra Nath Yogi 178

Contd…
• What is alpha-beta pruning procedure? Solve the following problem by using
this procedure

Contd…
• What are the different steps involved in simple problem solving?
• What is the main difference between Uninformed Search and
Informed Search strategies?
• What are the advantages and disadvantages of bidirectional
search strategy?
• What are the advantages of local search?

Thank You !
181By: Tekendra Nath Yogi12/25/2018

Unit3:Informed and Uninformed search

More Related Content

What's hot (20)

Similar to Unit3:Informed and Uninformed search (20)

More from Tekendra Nath Yogi (20)

Recently uploaded (20)

Unit3:Informed and Uninformed search