Download - Genetics Algorithms
-
8/3/2019 Genetics Algorithms
1/17
Seminar on
GENETIC ALGORITHMS
-
8/3/2019 Genetics Algorithms
2/17
CONTENTS
Introduction
Problem Formulation
Application of genetic algorithm
Results and Discussions
Conclusions
References
-
8/3/2019 Genetics Algorithms
3/17
Introduction
Bioinformatics is an application of computer technology.
Time table scheduling is allocation of various resources with
some constraints.
Resources of problem are rooms,lecturers,subjects,classes.
Two constraints are Soft constraints and Hard constraints.
Usually solved in academic organisations by harnessing the
power of computers.
Genetic algorithm makes it simpler.
-
8/3/2019 Genetics Algorithms
4/17
Problem Formulation
Problem definitionTimetabling problem can be defined as scheduling of resources
without conflicts
* Lecturer conflictsLecturer should not be assigned with two subjects of eithersame or different classes at the same time
* Subject conflictsTwo subjects should not be taught at the same time
* Room conflictsOne room should not be allocated to two classes at the
same time
-
8/3/2019 Genetics Algorithms
5/17
Time tablesClass time tables
Practical / lab time tables
Combined / complete time tables
Note: here assignment refers to
-
8/3/2019 Genetics Algorithms
6/17
Approches to automated time tabling using
genetic algorithm
Various versions:
Hybrid genetic algorithmParallel & distributed genetic algorithm
Genetic algorithm with divide & conquer technique
Genetic algorithm with greedy methods
-
8/3/2019 Genetics Algorithms
7/17
Fitness functionn m l r
Fitness value = (Cs(i)+Cl(j)+Cr(k))
p=0 i=0 j=0 k=0
Where Cs - subject conflictCl - lecturer conflict
Cr - room conflict
m - number of subjects
l - number of lecturers
r - number of rooms
n - number of classes
-
8/3/2019 Genetics Algorithms
8/17
Chromosome Representation
Binary Encoding
Value Encoding
Tree Encoding
Permutation Encoding
-
8/3/2019 Genetics Algorithms
9/17
Genetic operators
Crossover operator
Selection operator
Mutation operator
-
8/3/2019 Genetics Algorithms
10/17
Binary Encoding
11001011 + 11011101 = 11011111
Tree Encoding
-
8/3/2019 Genetics Algorithms
11/17
Application of GA to time table problem
Genetic algorithm:1. Initial population.
2. Apply GA process until optimum solution is obtained
Find fitness of all chromosomes.
Select parent fit for further processing. Apply crossover on selected parents.
Apply mutation on offspring generated in crossover.
Find fitness of mutated population.
Check for optimal result if not goto step2>.
3.End.
-
8/3/2019 Genetics Algorithms
12/17
Chromosome
representation scheme
Value encoding
Size of genetic material 5/6
Population size Variable(30 to 150)
Number of generations Variable(upto 2000)
Selection method Random with individual
selection method
Type of crossover Variable(set by user)
Crossover probability Variable(maximize for
better result)
Crossover point Variable(random
generation and program)
Type of mutation Alternate(inversion and
order changing)
Mutation probability Variable(min. for better
results)
Mutation point Variable(random
generation in program)
Coding scheme
-
8/3/2019 Genetics Algorithms
13/17
Figure4:Scheduler result for population size=150
Figure5:Comparisions of crossover types
Generation/Best Fitness values
-
8/3/2019 Genetics Algorithms
14/17
MERITS
1. Easy to apply for scheduling the timetables.
2. Overcomes the disadvantages of manual timetables.
3. Gives the appropriate results.
-
8/3/2019 Genetics Algorithms
15/17
DEMERITS
1. Should have the knowledge of the biological terms.
2. Good analytical skill is required.
-
8/3/2019 Genetics Algorithms
16/17
REFERENCES
Mahdi, Ainon, Roziati using a genetic
algorithm optimizer tool to generate good
quality timetables.
S.Sundaragopal, Rajani Boddu heuristic
optimization:complex university level
academic scheduling using GA. Goldberg D.E GA in search,optimization
& machine learning Addison Wesley.
-
8/3/2019 Genetics Algorithms
17/17
Conclusion
In this study we have implemented a genetictimetable scheduler by keeping the flexibility
to user.
A good schedule can be obtained by using
Pentium 4 with 256 MB of RAM.
The scheduler can be generalized for any large
problem.