1
Lahore Garrison University
CSC363-Artificial Intelligence
Week-6 Lecture-12
Semester#06
2
Preamble
Beyond Classical Search: Local Search
Local Beam Search
Keeps ‘k’ states for record and jump to best successor per need.
Stochastic Beam Search – ‘k’ random successors
Probability – Successor being an increasing function of its value.
Lahore Garrison University
3
Evolutionary Algorithm (Eas)
Genetic Algorithms (GAs) are search based algorithms based on the concepts of natural selection
and genetics.
GAs are a subset of a much larger branch of computation known as Evolutionary Computation.
Evolutionary algorithms have three main characteristics:
Population-Based The set of current solutions from which new solutions are to be generated is
called the population.
Fitness-Oriented There is a fitness value associated with each individual solution calculated from
a fitness function. Such fitness value reflects how good the solution is.
Variation-Driven individual solutions will undergo a number of variations to generate new
solutions.
Lahore Garrison University
4
Genetic Algorithm(John Holland)
GA is based on Darwin’s theory of evolution.
It is a slow gradual process that works by making changes to the making slight and slow
changes.
makes slight changes to its solutions slowly until getting the best solution.
This algorithm reflects the process of natural selection where the fittest individuals are
selected for reproduction in order to produce offspring of the next generation
Lahore Garrison University
5
Lahore Garrison University
6
Initial Population
The process begins with a set of individuals which is called a Population.
An individual is characterized by a set of parameters (variables) known as Genes.
Genes are joined into a string to form a Chromosome (solution).
Fitness Function
(also known as the Evaluation Function) evaluates how close a given solution is to the
optimum solution of the desired problem. It determines how fit a solution is
Lahore Garrison University
7
Selection
The idea of selection phase is to select the fittest individuals and let them pass their genes
to the next generation.
Two pairs of individuals (parents) are selected based on their fitness scores. Individuals
with high fitness have more chance to be selected for reproduction.
Crossover
also called recombination, is a genetic operator used to combine the genetic information of two
parents to generate new offspring. ...
Lahore Garrison University
8
Example (Crossover)
For example, consider the crossover point to be 3 as shown below:
Offspring are created by exchanging the genes of The new offspring are
parents among themselves until the crossover added to the population.
point is reached.
Lahore Garrison University
9
Mutation
In certain new offspring formed, some of their genes can be subjected to a mutation with a low
random probability. This implies that some of the bits in the bit string can be flipped.
Mutation occurs to maintain diversity within the population and prevent premature convergence.
Lahore Garrison University
10
Mutation
Note that without mutation the offspring will have all of its properties from its parents. To
add new features to such offspring, mutation took place
The individual after mutation is called mutant.
Lahore Garrison University
11
Genetic Algorithm
A variant of Stochastic Beam Search
Successor states are generated by combining two parent states.
Whereas, previous algorithms modified a single state.
Begins;
set of random ‘k’ states known as “Population”.
Each state is represented by a string over a finite alphabet.
A string of 0s and 1s.
Each state is known as an “Individual”.
Each state is rated by the Objective Function or “Fitness Function”.
Lahore Garrison University
12
Solving the 8-Queen problem using the
Genetic Algorithm
Step 1: Representing individuals
Step 2: Generating and initial Population
Step 3: Applying a Fitness Function
Step 4: Selecting parents for mating in accordance to their fitness
Step 5:Crossove of parents to produce new generation
Step 6: Mutation of new generation to bring diversity
Step 7: Repeat until solution is reached
Lahore Garrison University
13
Step 1: Representing individuals
Lahore Garrison University
14
Step 2: Generating and initial Population
Lahore Garrison University
15
Step 3: Applying a Fitness Function
Lahore Garrison University
16
Step 3: Applying a Fitness Function
Total = 23
Lahore Garrison University
17
Step 3: Applying a Fitness Function (Cont.)
Total =
78
Lahore Garrison University
18
Step 4: Selection
Lahore Garrison University
19
Step 4: Selection (Cont.)
Lahore Garrison University
20
Step 5: Crossover
Lahore Garrison University
21
Step 6: Mutation
Lahore Garrison University
22
Step 7: Repeat
Lahore Garrison University
23
Genetic Algorithm – Example 8-Queens
An 8-queens state must specify the positions of 8-Queens.
Each in a column of 8-squares.
Requires 8 x log2 8 = 24 bits.
State should be represented as 8-digits (1 to 8).
A fitness function should return higher values for better states.
We choose number of nonattacking pairs of queens.
Values of four states are 24, 23, 20 and 11 in current problem.
Probability of being chosen is directly proportional to Fitness score.
Lahore Garrison University
24
Genetic Algorithm - Example
Lahore Garrison University
25
Properties of similarity
Combine an uphill tendency with random exploration and exchange of information among
parallel search threads.
Advantage, genetic algorithm comes from the crossover operation.
If the positions of the genetic code are permuted initially in a random order, a crossover
conveys no advantage.
Raises the level of granularity.
A useful block can be combined with other blocks to construct a solution.
Lahore Garrison University
26
Algorithm
Lahore Garrison University
27
Lahore Garrison University
28
Lesson Preview
Beyond Classical Search: Local Search
Genetic Algorithm – A variant of Stochastic Beam Search
Generates successor states by combining two parent states.
Analogy to natural selection.
8-Queens problem
Initial population
Fitness function
Selection
Crossover
Mutation
Lahore Garrison University
29
Next Lesson Preview
Adversarial Search
Games
Optimal decisions in games
Alpha-Beta Pruning
Lahore Garrison University
30
References
To cover this topic, different reference material has been used for consultation.
Artificial Intelligence: A Modern Approach by Stuart J. Russell and Peter Norvig, 3 rd edition,
2010. Chapter # 4 – 4.1-4.3
Google search engine
Lahore Garrison University
