Date post: | 24-Dec-2015 |
Category: |
Documents |
Upload: | emory-hancock |
View: | 217 times |
Download: | 0 times |
11 / 76 / 76
Genetic AlgorithmsGenetic Algorithms
Authors:Authors:
Aleksandra Popovic, [email protected] Popovic, [email protected]
Aleksandra Jankovic, [email protected] Jankovic, [email protected]
Prof. Dr. Dusan Tosic, [email protected]. Dr. Dusan Tosic, [email protected]
Prof. Dr. Veljko Milutinovic, [email protected]. Dr. Veljko Milutinovic, [email protected]
22 / 76 / 76
Summary SlideSummary Slide
What You Will Learn From This Tutorial?What You Will Learn From This Tutorial?
33 / 76 / 76
What You Will Learn From This Tutorial?What You Will Learn From This Tutorial?
What is a genetic algorithm?What is a genetic algorithm? Principles of genetic algorithms.Principles of genetic algorithms. How to design an algorithm?How to design an algorithm? Comparison of gas and conventional algorithms.Comparison of gas and conventional algorithms.
Mathematics behind GA-sMathematics behind GA-s
Applications of GAApplications of GA– GA and the Internet GA and the Internet – Genetic search based on multiple mutation approaches Genetic search based on multiple mutation approaches
Part IIPart II
Part IPart I
Part IIIPart III
Part I: GA TheoryPart I: GA Theory
What are genetic algorithms?What are genetic algorithms?
How to design a genetic algorithm?How to design a genetic algorithm?
55 / 76 / 76
Genetic Algorithm Is Not...Genetic Algorithm Is Not...
...Gene...Gene codingcoding
66 / 76 / 76
Genetic Algorithm Is...Genetic Algorithm Is...
… … Computer algorithm Computer algorithm
That resides on principles of genetics and evolutionThat resides on principles of genetics and evolution
77 / 76 / 76
Instead of Introduction...Instead of Introduction...
Hill climbingHill climbing
locallocal
globalglobal
88 / 76 / 76
Instead of Introduction…(2)Instead of Introduction…(2)
Multi-climbersMulti-climbers
99 / 76 / 76
Instead of Introduction…(3)Instead of Introduction…(3) Genetic algorithmGenetic algorithm
I am not at the top.I am not at the top.My high is better!My high is better!
I am at the I am at the toptop
Height is ...Height is ...
I will continueI will continue
1010 / 76 / 76
Instead of Introduction…(3)Instead of Introduction…(3)
Genetic algorithm - few microseconds afterGenetic algorithm - few microseconds after
1111 / 76 / 76
GA ConceptGA Concept
Genetic algorithm (GA) introduces the principle of evolution Genetic algorithm (GA) introduces the principle of evolution and genetics into search among possible solutions and genetics into search among possible solutions to given problem.to given problem.
The idea is to simulate the process in natural systems. The idea is to simulate the process in natural systems. This is done by the creation within a machine This is done by the creation within a machine
of a population of individuals represented by chromosomes, of a population of individuals represented by chromosomes, in essence a set of character strings,in essence a set of character strings,that are analogous to the DNA,that are analogous to the DNA,that we have in our own chromosomes. that we have in our own chromosomes.
1212 / 76 / 76
Survival of the FittestSurvival of the Fittest
The main principle of evolution used in GA The main principle of evolution used in GA is “survival of the fittest”.is “survival of the fittest”.
The good solution survive, while bad ones die.The good solution survive, while bad ones die.
1313 / 76 / 76
Nature and GA...Nature and GA...
NatureNature Genetic algorithmGenetic algorithm
ChromosomeChromosome StringString
GeneGene CharacterCharacter
LocusLocus String positionString position
GenotypeGenotype PopulationPopulation
PhenotypePhenotype Decoded structureDecoded structure
1414 / 76 / 76
The History of GAThe History of GA
Cellular automata Cellular automata – John Holland, university of Michigan, 1975.John Holland, university of Michigan, 1975.
Until the early 80s, the concept was studied theoretically.Until the early 80s, the concept was studied theoretically. In 80s, the first “real world” GAs were designed.In 80s, the first “real world” GAs were designed.
1515 / 76 / 76
Algorithmic PhasesAlgorithmic Phases
Initialize the populationInitialize the population
Select individuals for the mating poolSelect individuals for the mating pool
Perform crossoverPerform crossover
Insert offspring into the populationInsert offspring into the population
The EndThe End
Perform mutationPerform mutation
yesyes
nono
Stop?Stop?
1616 / 76 / 76
Designing GA...Designing GA...
How to represent genomes?How to represent genomes? How to define the crossover operator?How to define the crossover operator? How to define the mutation operator?How to define the mutation operator? How to define fitness function?How to define fitness function? How to generate next generation?How to generate next generation? How to define stopping criteria?How to define stopping criteria?
1717 / 76 / 76
Representing Genomes...Representing Genomes...
RepresentationRepresentation ExampleExample
stringstring 1 0 1 1 1 0 0 11 0 1 1 1 0 0 1
array of stringsarray of strings http avala yubc net ~apopovichttp avala yubc net ~apopovic
tree - genetic programmingtree - genetic programming>>
bbxorxor
oror
cc
bbaa
1818 / 76 / 76
CrossoverCrossover
Crossover is concept from genetics.Crossover is concept from genetics. Crossover is sexual reproduction.Crossover is sexual reproduction. Crossover combines genetic material from two parents,Crossover combines genetic material from two parents,
in order to produce superior offspring.in order to produce superior offspring. Few types of crossover:Few types of crossover:
– One-pointOne-point– Multiple point.Multiple point.
1919 / 76 / 76
One-point CrossoverOne-point Crossover
Parent #1Parent #1 Parent #2Parent #2
00
11
55
33
55
44
77
66
77
66
22
44
22
33
00
11
2020 / 76 / 76
One-point CrossoverOne-point Crossover
Parent #1Parent #1 Parent #2Parent #2
00
11
55
33
55
44
77
66
77
66
22
44
22
33
00
11
2121 / 76 / 76
MutationMutation
Mutation introduces randomness into the population.Mutation introduces randomness into the population. Mutation is asexual reproduction.Mutation is asexual reproduction. The idea of mutation The idea of mutation
is to reintroduce divergence is to reintroduce divergence into a converging population.into a converging population.
Mutation is performed Mutation is performed on small part of population,on small part of population,in order to avoid entering unstable state.in order to avoid entering unstable state.
2222 / 76 / 76
Mutation...Mutation...
11 11 00 11 00 1100 00
00 11 00 11 00 1100 11
11 00
00 11
ParentParent
ChildChild
2323 / 76 / 76
About Probabilities...About Probabilities...
Average probability for individual to crossoverAverage probability for individual to crossoveris, in most cases, about 80%.is, in most cases, about 80%.
Average probability for individual to mutate Average probability for individual to mutate is about 1-2%.is about 1-2%.
Probability of genetic operators Probability of genetic operators follow the probability in natural systems.follow the probability in natural systems.
The better solutions reproduce more often.The better solutions reproduce more often.
2424 / 76 / 76
Fitness FunctionFitness Function
Fitness function is evaluation function,Fitness function is evaluation function,that determines what solutions are better than others.that determines what solutions are better than others.
Fitness is computed for each individual.Fitness is computed for each individual. Fitness function is application depended.Fitness function is application depended.
2525 / 76 / 76
SelectionSelection
The selection operation copies a single individual, The selection operation copies a single individual, probabilistically selected based on fitness, probabilistically selected based on fitness, into the next generation of the population. into the next generation of the population.
There are few possible ways to implement selection:There are few possible ways to implement selection:
– ““Only the strongest survive”Only the strongest survive”• Choose the individuals with the highest fitness Choose the individuals with the highest fitness
for next generationfor next generation
– ““Some weak solutions survive”Some weak solutions survive”• Assign a probability that a particular individual Assign a probability that a particular individual
will be selected for the next generationwill be selected for the next generation• More diversityMore diversity• Some bad solutions might have good parts!Some bad solutions might have good parts!
2626 / 76 / 76
Selection - Selection - Survival of The StrongestSurvival of The Strongest
0.930.93 0.510.51 0.720.72 0.310.31 0.120.12 0.640.64
Previous generationPrevious generation
Next generationNext generation
0.930.93 0.720.72 0.640.64
2727 / 76 / 76
Selection - Selection - Some Weak Solutions SurviveSome Weak Solutions Survive
0.930.93 0.510.51 0.720.72 0.310.31 0.120.12 0.640.64
Previous generationPrevious generation
Next generationNext generation
0.930.93 0.720.72 0.640.64 0.120.12
0.120.12
2828 / 76 / 76
Mutation and Selection...Mutation and Selection...
Phenotype
D
Phenotype
D
Phenotype
D
SelectionSelection MutationMutation
Solution distributionSolution distribution
2929 / 76 / 76
Stopping CriteriaStopping Criteria
Final problem is to decide Final problem is to decide when to stop execution of algorithm.when to stop execution of algorithm.
There are two possible solutions There are two possible solutions to this problem: to this problem: – First approach:First approach:
• Stop after production Stop after production of definite number of generationsof definite number of generations
– Second approach: Second approach: • Stop when the improvement in average fitness Stop when the improvement in average fitness
over two generations is below a thresholdover two generations is below a threshold
3030 / 76 / 76
GA Vs. Ad-hoc AlgorithmsGA Vs. Ad-hoc Algorithms
Genetic AlgorithmGenetic Algorithm Ad-hoc AlgorithmsAd-hoc Algorithms
SpeedSpeed
Human workHuman work
ApplicabilityApplicability
PerformancePerformance
SlowSlow * * Generally fastGenerally fast
MinimalMinimal Long and exhaustiveLong and exhaustive
GeneralGeneralThere are problems There are problems
that cannot be solved analyticallythat cannot be solved analytically
ExcellentExcellent DependsDepends
* Not necessary!* Not necessary!
3131 / 76 / 76
Problems With GasProblems With Gas
SometimesSometimes GA is extremely slow, GA is extremely slow, and much slower than usual algorithmsand much slower than usual algorithms
3232 / 76 / 76
Advantages of GasAdvantages of Gas
Concept is easy to understand.Concept is easy to understand. Minimum human involvement.Minimum human involvement. Computer is not learned how to use existing solution,Computer is not learned how to use existing solution,
but to find new solution!but to find new solution! Modular, separate from applicationModular, separate from application Supports multi-objective optimizationSupports multi-objective optimization Always an answer; answer gets better with time !!!Always an answer; answer gets better with time !!! Inherently parallel; easily distributedInherently parallel; easily distributed Many ways to speed up and improve a GA-based application as Many ways to speed up and improve a GA-based application as
knowledge about problem domain is gainedknowledge about problem domain is gained Easy to exploit previous or alternate solutionsEasy to exploit previous or alternate solutions
3333 / 76 / 76
GA: An Example -GA: An Example - Diophantine EquationsDiophantine Equations
Diophantine equation (n=4):Diophantine equation (n=4):
A*x + b*y + c*z + d*q = sA*x + b*y + c*z + d*q = s
For given a, b, c, d, and s - find x, y, z, qFor given a, b, c, d, and s - find x, y, z, q
Genome:Genome:
(X, y, z, p) = (X, y, z, p) =
xx yy zz qq
3434 / 76 / 76
GA:An Example -GA:An Example - Diophantine Equations(2)Diophantine Equations(2)
CrossoverCrossover
MutationMutation
( 1, 2, 3, 4 )( 1, 2, 3, 4 )
( 5, 6, 7, 8 )( 5, 6, 7, 8 )
( 1, 6, 3, 4 )( 1, 6, 3, 4 )
( 5, 2, 7, 8 )( 5, 2, 7, 8 )
( 1, 2, 3, 4 )( 1, 2, 3, 4 ) ( 1, 2, 3, 9 )( 1, 2, 3, 9 )
3535 / 76 / 76
GA:An Example -GA:An Example - Diophantine Equations(3)Diophantine Equations(3)
First generation is randomly generated of numbers First generation is randomly generated of numbers lower than sum (s).lower than sum (s).
Fitness is defined as absolute value of difference Fitness is defined as absolute value of difference between total and given sum:between total and given sum:
Fitness = abs ( total - sum ) ,Fitness = abs ( total - sum ) ,
Algorithm enters a loop in which operators are performed Algorithm enters a loop in which operators are performed on genomes: crossover, mutation, selection.on genomes: crossover, mutation, selection.
After number of generation a solution is reached.After number of generation a solution is reached.
Part II: Part II: Mathematics Behind GA-sMathematics Behind GA-s
Two methods for analyzing genetics algorithms:Two methods for analyzing genetics algorithms:
Schema analysesSchema analyses
Mathematical modelingMathematical modeling
3737 / 76 / 76
Schema AnalysesSchema Analyses
Weaknesses:Weaknesses: In determining some characteristics of the populationIn determining some characteristics of the population Schema analyses makes some approximations that weaken itSchema analyses makes some approximations that weaken it
Advantages:Advantages: A simple way to view the standard GAA simple way to view the standard GA They have made possible proofs of some interesting theoremsThey have made possible proofs of some interesting theorems They provide a nice introduction to algorithmic analysesThey provide a nice introduction to algorithmic analyses
3838 / 76 / 76
Schema Analyses…Schema Analyses…
Schema – a template made up of a string of 1s, 0s, and *s, Schema – a template made up of a string of 1s, 0s, and *s,
where * is used as a wild card that can be either 1 or 0 where * is used as a wild card that can be either 1 or 0
For example, For example, H = 1 * * 0 * 0H = 1 * * 0 * 0 is a schema. is a schema. It has eight It has eight instancesinstances (one of which is 101010) (one of which is 101010) OrderOrder, , o ( H )o ( H ) , the number of non-*, or defined, bits (in example 3), the number of non-*, or defined, bits (in example 3) Defining lengthDefining length, , d ( H )d ( H ) , greatest distance between two defined bits , greatest distance between two defined bits
(in example H has a defining length of 3) (in example H has a defining length of 3)
Let S be the set of all strings of length Let S be the set of all strings of length l.l. There is possible schemas on S, but different subsets of SThere is possible schemas on S, but different subsets of S Schema cannot be used to represent every possible population within S, Schema cannot be used to represent every possible population within S,
but forms a representative subset of the set of all subsets of Sbut forms a representative subset of the set of all subsets of S
l3l22
3939 / 76 / 76
Schema analyses…Schema analyses…
The end-of-iterations conditions:The end-of-iterations conditions:
expected number of instances of schema H as we iterate the GAexpected number of instances of schema H as we iterate the GA M(H, t) – the number of instances of H at time tM(H, t) – the number of instances of H at time t f(x) – fitness of chromosome xf(x) – fitness of chromosome x - average fitness at time t :- average fitness at time t :
, n=|S|, n=|S|
- average fitness of instances of H at time t:- average fitness of instances of H at time t:
)(tf
n
xftf Sx
)()(
),(ˆ tHu
),(
)(),(ˆ
tHm
xftHu Hx
4040 / 76 / 76
Schema Analyses…Schema Analyses…
- If we completely ignore the effects of crossover and mutation, If we completely ignore the effects of crossover and mutation,
we get the expected value:we get the expected value:
- Now we consider only the effects of crossover and mutation,Now we consider only the effects of crossover and mutation,
which lower the number of instances of H in the population.which lower the number of instances of H in the population.
Then we will get a good lower bound on E(m(H,t+1))Then we will get a good lower bound on E(m(H,t+1)) - probability that a random crossover bit is between the - probability that a random crossover bit is between the
defining bits of H defining bits of H - probability of crossover occurring- probability of crossover occurring
)(
),(),(ˆ
)(
)(
)(
)())1,((
tf
tHmtHu
tf
xf
xf
xfntHmE Hx
Sx
Hx
)(HSc
cp)
1
)((1)(
l
HdpHS cc
4141 / 76 / 76
Schema Analyses…Schema Analyses…
- probability of an instance of H remaining the same after - probability of an instance of H remaining the same after mutation; it is dependent on the order of H mutation; it is dependent on the order of H
- probability of mutation- probability of mutation
- With the above notation , we have:With the above notation , we have:
Schema TheoremSchema Theorem, provided by John Holland , provided by John Holland
)(HSm
mp)()1()( Ho
mm pHS
))1)((1
)(1(
)(
),(),(ˆ))1,(( )(Ho
mc pl
Hdp
tf
tHmtHutHmE
4242 / 76 / 76
Schema analyses…Schema analyses…
The Schema Theorem only shows how schemas The Schema Theorem only shows how schemas dynamically change, and how short, low-order schemas whose dynamically change, and how short, low-order schemas whose fitness remain above the average mean receive fitness remain above the average mean receive
exponentially growing increases in the number of samples. exponentially growing increases in the number of samples.
It cannot make more direct predictions about the It cannot make more direct predictions about the population composition, distribution of fitness and other statistics population composition, distribution of fitness and other statistics more directly related to the GA itself.more directly related to the GA itself.
4343 / 76 / 76
Mathematical ModelMathematical Model One change is that one of the new individuals One change is that one of the new individuals
will be immediately deleted (thus the loop is iteratedwill be immediately deleted (thus the loop is iterated n n times, not times, not n/2n/2)) S – set of all strings of length lS – set of all strings of length l N – size of S, or N – size of S, or
column vector with rows such that thecolumn vector with rows such that the i- i-th componentth component
is equal to the proportion of the population P as timeis equal to the proportion of the population P as time t t that has chromosome that has chromosome i i
column vector with rows such thatcolumn vector with rows such that i i- th component - th component
is equal to the probability that chromosome i will be selected as a parentis equal to the probability that chromosome i will be selected as a parent
)(tp )(tpi
)(ts
)(tsi
4444 / 76 / 76
Mathematical Model…Mathematical Model…
ExampleExample:: l=2, 3 l=2, 3 individuals in the population, two with individuals in the population, two with chromosome 10 and one with chromosome 11, then chromosome 10 and one with chromosome 11, then
If the fitness is equal to the number of 1s in the string, thenIf the fitness is equal to the number of 1s in the string, then
Ttp )3
1,
3
2,0,0()(
Tts )2
1,
2
1,0,0()(
4545 / 76 / 76
Mathematical Model…Mathematical Model…
diagonal matrix with diagonal matrix with
Relation between and : Relation between and :
Goal: given a column vector , to construct a column-vector-valued Goal: given a column vector , to construct a column-vector-valued function such thatfunction such that
M M represents recombination – composition of crossover and mutation represents recombination – composition of crossover and mutation component wise sum of i and j mod 2component wise sum of i and j mod 2
component wise product of i and jcomponent wise product of i and j
matrix whose i , j th entry is the probability that 0 matrix whose i , j th entry is the probability that 0 result from the recombination of i and jresult from the recombination of i and j
NN)( , jiFF )(, ifF ii
p s
)(
)()(tp
tp
F
Fts
x
)(xM
)1())(( tptsΜ
ji
ji
)( ,jiMM jiM ,
4646 / 76 / 76
Mathematical Model…Mathematical Model…
permutation operator on permutation operator on
Finally, is given by the following expressionFinally, is given by the following expression
With this expression, we can calculate explicitly the expected value of each With this expression, we can calculate explicitly the expected value of each generation from the proceeding generation.generation from the proceeding generation.
I hope you now fully understand the mathematics behind GA.I hope you now fully understand the mathematics behind GA.
j NR
TNjj
TNoj yyyy ),...,()),...,(( )1(01
)(sM
)))((,...,))((()( )(12)(0120ts
Tts
Tl
l tstssM
MM
Part III: Applications of GAsPart III: Applications of GAs
GA and the InternetGA and the Internet
Genetic search based on multiple mutation approachesGenetic search based on multiple mutation approaches
4848 / 76 / 76
Some Applications of GasSome Applications of Gas
GAGAInternet searchInternet search
Data miningData mining
Software guided circuit designSoftware guided circuit designControl systems designControl systems design
Stock prize predictionStock prize prediction
Path findingPath finding Mobile robotsMobile robotssearchsearch
OptimizationOptimization
Trend spottingTrend spotting
Genetic Algorithm Genetic Algorithm and the Internet and the Internet
The system designed by EBI Group,The system designed by EBI Group,Faculty for Electrical Engineering,Faculty for Electrical Engineering,
University of BelgradeUniversity of Belgrade
5050 / 76 / 76
Algorithms PhasesAlgorithms Phases
Process set of URLs given by userProcess set of URLs given by user
Select all links from input setSelect all links from input set
Evaluate fitness function for all genomesEvaluate fitness function for all genomes
Perform crossover, mutation, and reproductionPerform crossover, mutation, and reproduction
SatisfactorySatisfactorysolutionsolution
obtained?obtained?
The EndThe End
5151 / 76 / 76
IntroductionIntroduction
GA can be used for intelligent internet search.GA can be used for intelligent internet search. GA is used in cases when search space GA is used in cases when search space
is relatively large.is relatively large. GA is adoptive search.GA is adoptive search. GA is heuristic search method.GA is heuristic search method.
5252 / 76 / 76
System for GA Internet System for GA Internet SearchSearch Designed at faculty for electrical engineering, university of belgradeDesigned at faculty for electrical engineering, university of belgrade
CCOONNTTRROOLL
PPRROOGGRRAAMM
AgentAgent SpiderSpider
Input setInput set
TopicTopic
SpaceSpace
TimeTime
Output setOutput set
Current setCurrent set
Top dataTop data
Net dataNet data
GeneratorGenerator
5353 / 76 / 76
SpiderSpider
Spider is software packages,Spider is software packages, that picks up internet documents that picks up internet documents from user supplied input with depth specified by user. from user supplied input with depth specified by user.
Spider takes one URL, fetches all links, Spider takes one URL, fetches all links, and documents thy contain with predefined depth. and documents thy contain with predefined depth.
The fetched documents are stored on local hard disk with same The fetched documents are stored on local hard disk with same structure as on the original location.structure as on the original location.
Spider’s task is to produce the first generation.Spider’s task is to produce the first generation. Spider is used during crossover and mutation.Spider is used during crossover and mutation.
5454 / 76 / 76
AgentAgent
Agent takes as an input a set of urls, Agent takes as an input a set of urls, and calls spider, for every one of them, with depth 1. and calls spider, for every one of them, with depth 1.
Then, agent performs extraction of keywords Then, agent performs extraction of keywords from each document, and stores it in local hard disk.from each document, and stores it in local hard disk.
5555 / 76 / 76
GeneratorGenerator
Generator generates a set of urls from given keywords, Generator generates a set of urls from given keywords, using some conventional search engine. using some conventional search engine.
It takes as input the desired topic, calls yahoo search engine, It takes as input the desired topic, calls yahoo search engine, and submits a query looking for all documents and submits a query looking for all documents covering the specific topic. covering the specific topic.
Generator stores URL and topic of given web page Generator stores URL and topic of given web page in database called topdata.in database called topdata.
5656 / 76 / 76
TopicTopic
It uses topdata DB in It uses topdata DB in order to insert random urls order to insert random urls from database into current set.from database into current set.
Topic performs mutation.Topic performs mutation.
5757 / 76 / 76
SpaceSpace
Space takes as input the current set Space takes as input the current set from the agent application from the agent application and injects into it those urls and injects into it those urls from the database netdata from the database netdata that appeared with the greatest frequency that appeared with the greatest frequency in the output set of previous searches.in the output set of previous searches.
5858 / 76 / 76
TimeTime
Time takes set of urls from agent Time takes set of urls from agent and inserts ones with greatest frequency into DB netdata.and inserts ones with greatest frequency into DB netdata.
The netdata DB contains of three fields: URL, topic, The netdata DB contains of three fields: URL, topic, and count number. and count number.
The DB is updated in each algorithm iteration.The DB is updated in each algorithm iteration.
5959 / 76 / 76
How Does The System Work?How Does The System Work?
CCOONNTTRROOLL
PPRROOGGRRAAMM
AgentAgent SpiderSpider
Input setInput set
TopicTopic
SpaceSpace
TimeTime
Output setOutput set
Current setCurrent set
Top dataTop data
Net dataNet data
GeneratorGenerator
command flowcommand flow
data flowdata flow
6060 / 76 / 76
GA and the Internet: ConclusionGA and the Internet: Conclusion
GA for internet search, on contrary to other gas,GA for internet search, on contrary to other gas,is much faster and more efficient that conventional solutions,is much faster and more efficient that conventional solutions,such as standard internet search engines.such as standard internet search engines.
INTERNETINTERNET
6161 / 76 / 76
Genetic Search Based on Genetic Search Based on Multiple Mutation ApproachesMultiple Mutation Approaches
Concept and its improvements adapted to specific Concept and its improvements adapted to specific applications in e-business, and concrete software packageapplications in e-business, and concrete software package
Main problems in finding information on the Internet:Main problems in finding information on the Internet: How to find quickly and retrieve efficiently the potentially useful How to find quickly and retrieve efficiently the potentially useful
information considering the fact of the fast growth of the quantity information considering the fact of the fast growth of the quantity and variety of Internet sitesand variety of Internet sites
Huge number of documents , many of which are completely Huge number of documents , many of which are completely unrelated to what the user originally attempted to find, searched unrelated to what the user originally attempted to find, searched with indexing engineswith indexing engines
Documents placed on the top of the result list are often less Documents placed on the top of the result list are often less acceptable then the lower onesacceptable then the lower ones
Indexing process may take days, weeks , or even longer, because Indexing process may take days, weeks , or even longer, because the volume of new information being created dailythe volume of new information being created daily
6262 / 76 / 76
Links Based ApproachLinks Based Approach
The The questionquestion is: is:
How to locate and retrieve the needed information before it gets indexed?How to locate and retrieve the needed information before it gets indexed?
TheThe efficient way efficient way to locate the new not-yet-indexed information:to locate the new not-yet-indexed information: Using links-based approaches genetic searchUsing links-based approaches genetic search
simulated annealingsimulated annealing
Best Best result:result:
indexing - based approachesindexing - based approaches
++ links - based approacheslinks - based approaches
6363 / 76 / 76
Genetic Search AlgorithmGenetic Search Algorithm
GENETIC ALGORITHM OF ZERO ORDER, with no mutationGENETIC ALGORITHM OF ZERO ORDER, with no mutation
Start:Start:Model Web presentation that contains all the needed types of Model Web presentation that contains all the needed types of information (fitness function is evaluated).information (fitness function is evaluated).
It is assumes that it includes URL pointers to other similar Web It is assumes that it includes URL pointers to other similar Web presentations, and these are downloaded.presentations, and these are downloaded.The Web presentations that “survived” the fitness function are The Web presentations that “survived” the fitness function are assumed to include additional URL pointers, and their related Web assumed to include additional URL pointers, and their related Web presentations are downloaded next.presentations are downloaded next.After the end-of-search condition is met, the Web presentations are After the end-of-search condition is met, the Web presentations are ranked according to their fitness value.ranked according to their fitness value.
6464 / 76 / 76
Genetic Search Algorithm…Genetic Search Algorithm…
Type of mutation:Type of mutation:
Topic-oriented database mutationTopic-oriented database mutation
Semantic mutationsSemantic mutations
- based on the principles of spatial locality- based on the principles of spatial locality
- based on the principles of temporal locality- based on the principles of temporal locality
Logical reasoning and semantics consideration is involve in Logical reasoning and semantics consideration is involve in picking out URLs for mutation.picking out URLs for mutation.
6565 / 76 / 76
Innovations Required by Domain AreaInnovations Required by Domain Area
APPLICATION LEVELAPPLICATION LEVEL
LEVEL OF THE GENERAL PROJECT APPROACHLEVEL OF THE GENERAL PROJECT APPROACH
AND PRODUCT ARCHITECTUREAND PRODUCT ARCHITECTURE
ALGORITHMIC LEVELALGORITHMIC LEVEL
IMPLEMENTATION LEVEL IMPLEMENTATION LEVEL
6666 / 76 / 76
Application LevelApplication Level
Statistical analysis and data mining has to be performed,Statistical analysis and data mining has to be performed,
in order to figure out the common and typical patterns of in order to figure out the common and typical patterns of behavior and needbehavior and need
The state-of-the-art of mutual referencing has to be determinedThe state-of-the-art of mutual referencing has to be determined
The trends and asymptotic situations foreseen for the time of The trends and asymptotic situations foreseen for the time of project finalization has to be determinedproject finalization has to be determined
6767 / 76 / 76
Level of the General Project Level of the General Project Approach and Product ArchitectureApproach and Product Architecture
Decisions have to be made about the most important goals to be Decisions have to be made about the most important goals to be achieved:achieved:
Maximizing the speed of searchMaximizing the speed of search
Maximizing the sophistication of searchMaximizing the sophistication of search
Maximizing specific effects of interest for a given institution or a Maximizing specific effects of interest for a given institution or a customercustomer
Maximizing a combination of the aboveMaximizing a combination of the above
Decision on this level affect the applicability of the final product / Decision on this level affect the applicability of the final product / tool.tool.
6868 / 76 / 76
Algorithmic LevelAlgorithmic Level
Develop an efficient mutation algorithm of interest for the applicationDevelop an efficient mutation algorithm of interest for the application
in the direction of database architecture and designin the direction of database architecture and design
in introducing the elements of semantic-based mutationin introducing the elements of semantic-based mutation
Semantics-based mutations are especially of interest for chaotic Semantics-based mutations are especially of interest for chaotic markets, typical of new markets in developed countries or markets, typical of new markets in developed countries or traditional markets in under-developed countries.traditional markets in under-developed countries.
6969 / 76 / 76
Semantics-based MutationSemantics-based Mutation
Mutation based on spatial localitiesMutation based on spatial localities
After a “fruitful” Web presentation is reached (using a tradicional After a “fruitful” Web presentation is reached (using a tradicional algorithm with mutation), the site of the same Internet service provider algorithm with mutation), the site of the same Internet service provider is searched for other presentations on the same or similar topicis searched for other presentations on the same or similar topic
Explanation :Explanation :
In chaotic markets, it is very unlikely that service/product offers from In chaotic markets, it is very unlikely that service/product offers from the same small geographic area each other on their Web presentationsthe same small geographic area each other on their Web presentations
After a successful “side trip” based on spatial mutation, one continue After a successful “side trip” based on spatial mutation, one continue with the traditional database mutation.with the traditional database mutation.
7070 / 76 / 76
Semantics-based Mutation…Semantics-based Mutation…
Mutation based on temporal localitiesMutation based on temporal localities
One comes back periodically to a Web presentation which was One comes back periodically to a Web presentation which was “fruitful” in the past“fruitful” in the past
One comes back periodically to other Web presentations One comes back periodically to other Web presentations developed by the author who created some “fruitful” Web developed by the author who created some “fruitful” Web presentations in the pastpresentations in the past
Temporal mutation can use direct revisits or a number of indirect Temporal mutation can use direct revisits or a number of indirect forms or revisit.forms or revisit.
7171 / 76 / 76
Implementation LevelImplementation Level
Utilization of novel technologies, for maximal performance and Utilization of novel technologies, for maximal performance and minimal implementation complexityminimal implementation complexity
Important for:Important for:
- good flexibility- good flexibility
- extendibility- extendibility
- reliability- reliability
- availability- availability Utilization of mobile platforms and mobile agentsUtilization of mobile platforms and mobile agents
7272 / 76 / 76
Implementation Level…Implementation Level… Static agentsStatic agents
- one has to download megabytes of information- one has to download megabytes of information- treat that information with a decision-making code of size - treat that information with a decision-making code of size measured in kilobytesmeasured in kilobytes- derive the final business related decision, which is binary in size - derive the final business related decision, which is binary in size (one bit: yes or no)(one bit: yes or no)
A huge amount of data is transferred through the network A huge amount of data is transferred through the network in vain, because only a small percent of fetched documents will in vain, because only a small percent of fetched documents will turn out to be usefulturn out to be useful
Mobile agentsMobile agents- they would browse through the network and perform the search - they would browse through the network and perform the search locally, on the remote servers, transferring only the needed locally, on the remote servers, transferring only the needed documents and datadocuments and data- they load the network only with kilobytes and a single bit- they load the network only with kilobytes and a single bit
7373 / 76 / 76
Simulation ResultSimulation Result
Links-based approach in the static domainLinks-based approach in the static domain How various mutation strategies can affect the search efficiencyHow various mutation strategies can affect the search efficiency Set of software packages have developed , that would perform Set of software packages have developed , that would perform
Internet search using genetic algorithms (by Veljko Milutinovic, Internet search using genetic algorithms (by Veljko Milutinovic, Dragana Cvetkovic, and Jelena Mirkovic)Dragana Cvetkovic, and Jelena Mirkovic)
As the fitness function they have measured average Jaccard’s As the fitness function they have measured average Jaccard’s score for the output documents, while changing the type and rate score for the output documents, while changing the type and rate of mutationof mutation
7474 / 76 / 76
Simulation Result…Simulation Result…
The simulation result for topic The simulation result for topic mutationmutation
The simulation result for temporal The simulation result for temporal and spatial mutation combined and spatial mutation combined with topic mutationwith topic mutation
7575 / 76 / 76
Simulation Result…Simulation Result…
The simulation result for topic, The simulation result for topic, spatial and temporal mutation spatial and temporal mutation combined.combined.
Constant increase in the quality Constant increase in the quality of pages found.of pages found.
7676 / 76 / 76
Conclusion: EvolutionConclusion: Evolution
Tutorial download: galeb.etf.bg.ac.yu/~vm Option:TutorialsTutorial download: galeb.etf.bg.ac.yu/~vm Option:Tutorials