Artificial Intelligence

Lecture – 10
Prepared By: Prof (Dr.) Anirban Bose
Dept.: Computer Science & Engineering
Dr. B. C. Roy Engineering College, Durgapur
2 Simulated Annealing
 In metallurgy, annealing is the process used to temper or harden metals and glass by
heating them to a high temperature and then gradually cooling them, thus allowing the
material to reach a low energy crystalline state.

 The loop of the simulated-annealing algorithm is quite similar to hill climbing. Instead of
picking the best move, however, it picks a random move.
 If the move improves the situation, it is always accepted. Otherwise, the algorithm accepts
the move with some probability less than 1.
3 An Example
Initial State – S
1 2 3
Goal State – G
4 5 Cost/Heuristic Function, E(n) – No. of Misplaced Tiles
6 7 8 Temperature, T – 100
Cooling Schedule – 0.95
Start (S) Transition Function – Function to generate a neighbouring
state from the current state.
Acceptance Probability, P -
DeltaE,E – E(n’) – E(n)

1 2 3 Algo:
while(iter <= MaxIter)
4 5 6
while(T > Tmin )
7 8 n’ = generateNeighbour(n)
deltaE(n, n’, T)
Goal (G)
if(deltaE <= 0 or r <= P)
n n’
T T*0.95
iter iter + 1
4 Simulated Annealing…

 The probability decreases exponentially with the “badness” of the move—the amount ΔE
by which the evaluation is worsened.

 The probability also decreases as the “temperature” T goes down: “bad” moves are more
likely to be allowed at the start when T is high, and they become more unlikely as T
decreases.

 If the schedule (mapping of time to temperature and how fast the temperature decreases is
called the Annealing Schedule) lowers T slowly enough, the algorithm will find a global
optimum with probability approaching 1.
5 Genetic Algorithms
 A genetic algorithm (or GA) is a variant of stochastic beam search in which successor
states are generated by combining two parent states rather than by modifying a single state.
 Like beam searches, GAs begin with a set of k randomly generated states, called the
population.
 Each state, or individual, is represented as a string over a finite alphabet—most commonly,
a string of 0s and 1s. For example, an 8-queens state must specify the positions of 8
queens, each in a column of 8 squares, and so requires 8 × log 2 8 = 24 bits.
 Alternatively, the state could be represented as 8 digits, each in the range from 1 to 8.
 Each state is rated by the objective function, or (in GA terminology) the fitness function.
 A fitness function should return higher values for better states.
6 How GA Works?
7 Operators in GA

 Selection: It is one of the most popular ways of parent selection. In this every individual
can become a parent with a probability which is proportional to its fitness.

 Crossover: The crossover operator is analogous to reproduction and biological crossover.

In this more than one parent is selected and one or more off-springs are produced using the
genetic material of the parents. Crossover is usually applied in a GA with a high
probability.

 Mutation: It may be defined as a small random tweak in the chromosome, to get a new
solution. It is used to maintain and introduce diversity in the genetic population and is
usually applied with a low probability.
8 Operators in Action
Fitness Function: X2 (Minimization)

1 0 1 0 0 1 1 1 1 0 1 0 0 1 0 0
pa1 off1

0 0 0 1 1 1 0 0 0 0 0 1 1 1 1 1
pa2 off2
Crossover point
pa1 fitness value: 1672=27889 off1 fitness value: 26896
pa2 fitness value = 784 off2 fitness value = 961

1 0 1 0 0 1 0 0 1 0 1 0 0 0 0 0
off1 off1 off1 fitness value: 25600
off2 fitness value = 729
0 0 0 1 1 1 1 1 0 0 0 1 1 0 1 1
off2 Mutation point off2
9 8 - Queens
 Problem: placing eight queens on an 8×8
chessboard such that none of them attack one
another (no two are in the same row, column,
or diagonal).
 States: All configuration of the chessboard
with queens.
 State Transition Model: Moving a queen in
distinct places within a column.
 Fitness Function: Number of pairs of non-
attacking queens in a particular state.
10 8 - Queens

 Genetic Algorithms do not require any derivative information.

 Genetic Algorithms are faster and efficient when compared to the traditional methods of
brute-force search.
 Genetic Algorithms is proven to have many parallel capabilities
 Optimizes both continuous and discrete functions and also multi-objective problems.
11 Constraint Satisfaction Problem
 A constraint satisfaction problem consists of three components, X, D, and C:
 X is a set of variables, {X1, . . . , Xn}.
 D is a set of domains, {D1, . . . , Dn}, one for each variable.
 C is a set of constraints that specify allowable combinations of values.
 Each domain Di consists of a set of allowable values, {v1, . . . , vk} for variable Xi. Each
constraint Ci consists of a pair <scope, rel>, where ‘scope’ is a tuple of variables that participate
in the constraint and ‘rel’ is a relation that defines the values that those variables can take on.
 To solve a CSP, we need to define a state space and the notion of a solution. Each state in a CSP
is defined by an assignment of values to some or all of the variables, {X i = vi, Xj = vj, . . .}.
 An assignment that does not violate any constraints is called a consistent or legal assignment.
 A complete assignment is one in which every variable is assigned, and a solution to a CSP is a
consistent, complete assignment.
 A partial assignment is one that assigns values to only some of the variables.
12 Example:
 Suppose that, we are looking at a map of Australia showing each of its states and territories. We are
given the task of coloring each region either red, green, or blue in such a way that no neighboring
regions have the same color.

WA: Western Australia

NT: Northern Territory
SA: South Australia
Q: Queensland
NSW: New South Wales
V: Victoria
T: Tasmania

 To formulate this as a CSP, we define the variables to be the regions X = {WA,NT, Q,NSW , V,SA, T }.
The domain of each variable is the set Di = {red,green,blue}. The constraints require neighboring
regions to have distinct colors. Thus there are nine constraints: C = {SA WA,SA NT,SA Q,SA
NSW ,SA V, WA NT,NT Q, Q NSW ,NSW V }
13 Example…
 Variables, .
 Domains, .
 Constraints,
OR

 The family of CSP constraints can be better visualized by Constraint Graph.

• The nodes of the graph correspond to variables of the problem, and a link
connects any two variables that participate in a constraint.
• CSP solvers can be faster than state-space searchers because the CSP solver
can quickly eliminate large swatches of the search space.
• For example, once we have chosen {SA = blue}, a search procedure would
have to consider 35 = 243 assignments for the five neighboring variables
instead of only 25 = 32 assignments to look at, a reduction of 87%.
14 Types of CSP Variables
■ Discrete variables
■ finite domains:
■ n variables, domain size d  O (d n) complete assignments

■ e.g., map coloring and 8-queens problems are both of this kind.


infinite domains:
■ integers, strings, etc.
■ e.g., job scheduling, variables are start/end days for each job

■ need a constraint language, e.g., StartJob + 5 ≤ StartJob

1 3
■ Continuous variables
■ e.g., start/end times for Hubble Space Telescope observations
■ linear constraints solvable in polynomial time by linear programming
15 Types of Constraints
 Unary constraints involve a single variable.
 e.g., SA ≠ green
 Binary constraints involve pairs of variables.
 e.g., SA ≠ WA
 Higher-order constraints involve 3 or more variables.
 e.g., cryptarithmetic column constraints
 A constraint involving an arbitrary number of variables is called a global constraint. One of
the most common global constraints is Alldiff , which says that all of the variables involved
in the constraint must have different values.
 e.g., cryptarithmetic puzzles, 8 – queen.
16 Solving CSP: Backtracking
 Only need to consider assignments to a single variable at each node.
 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.
 Backtracking search is the basic uninformed algorithm for CSPs
17 Map Coloring Problem

WA NT Q NSW V SA T
Initial Domain RGB RGB RGB RGB RGB RGB RGB
After WA = Red R GB RGB RGB RGB GB RGB
After Q = Green R B G R B RGB B RGB
After V = Blue R B G R B ??? RGB
18 Map Coloring Problem

WA NT Q NSW V SA T
Initial Domain RGB RGB RGB RGB RGB RGB RGB

After SA = Blue RG RG RG RG RG B RGB

After WA = Red R G RG RG RG B RGB

After NT = Green R G R G R B G
19 Map Coloring Problem: Backtrack
20 LOCAL SEARCH FOR CSPS
 Local search algorithms (Hill Climbing, Simulated Annealing etc.) turn out to be effective in solving many
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.
 Typically, the initial guess violates several constraints. The point of local search is to eliminate the violated
constraints.
 In choosing a new value for a variable, the most obvious heuristic is to select the value that results in the
minimum number of conflicts with other variables
21 Example
22 Excercise

X = {V1, V2, V3, V4, V5, V6}

D = {R, G, B}
C = If there is an edge
between i and j, then Vi
must be different of Vj