Evolutionary Computing
Lecture 1(Introduction)
Bu-Ali Sina UniversityComputer Engineering Dep.
Fall 2010
Outline
� Syllabus� References� Course Plane� Grading and policies� Introduction to Evolutionary Computing
Introduction
Syllabus
MSRT References
References
Recommended journals
�IEEE tran. On Evolutionary Computation
�International journal of Applied EvolutionaryComputation (IJAEC)
� Evolutionary Computation
�Genetic programming and evolvable machines
�Journal of artificial evolution and applications
EC Conferences
•IEEE wcci (Paper sub. 31 jan.
•Genetic and evolutionary computation conference (GECCO)
•Congress on evolutionary computation (CEC)
•evolutionary programming (EP)
•Parallel problem solving from nature (PPSN)
•EvoStar 2010 (paper sub 30 Nov.)
•UC10 (paper sub 1 feb)
•Alife ( 28 feb.)
•Kes (1 mar. )
•ICES (5 mar.)
•CEC
Course plan• Introduction• Evolutionary Algorithm• Genetic Algorithm• Evolution Strategy• Evolutionary Programming• Genetic Programming• Parameter Control• Multi-modal Problems and Spatial Distribution• Hybrid Evolutionary Algorithms
Grading and PoliciesExams 50%
– Midterm 50% (25% of total)– Final 50% (25% of total)
Final Project (25%)– One project (deadline is about 31/4/90)
Seminar (15%)– Every body present a seminar (select a subject until
15/8/90)Home works (10%)
– 5 home works
Hours: Mon. 10-13
Site: http://Profs.basu.ac.ir/khotanlou
Email:[email protected]
Contact: 8257410, 11 (324 )
This Lecture Outline• The basic EC metaphor• Historical perspective• Biological inspiration:
– Darwinian evolution theory– Genetics– Motivation for EC
• What can EC do: examples of applicationareas
Evolutionary Computation• Elements of Evolution:
– Reproduction– Random variation– Competition– Selection of contending individuals from apopulation.
• Evolutionary computation:computational methods simulating evolution,
mostly used to find a solution in a large searchspace.
Optimization
oEnvironment of an organism and its survival chance in theenvironment vs. evaluation of parameter to optimize for asolution candidate.
o Start from a random sample of solution candidates andsimulate natural evolution, optimizing an evaluation function(fitness of the individual).
o Classical methods: gradient descent, deterministic hillclimbing, random search.
o problems: nonlinear, stochastic, temporal, or with multiplelocal optima.
Machine Intelligence
•Capability of a system to adapt its behavior tomeet desired goals in a range of environments.
• Evolution of organisms » natural intelligence
•Evolutionary computation can be used to evolvethe data in an artificial intelligence model.
Biology
•Using computation to simulate the evolutionand understand the evolution of organisms.
• Rather using computation in biology thensimulating biological evolution for computation.
History•The idea of using simulated evolution to solveengineering and design problems have been aroundsince the 1950’s.
– Bremermann, 1962– Box, 1957– Friedberg, 1958
• However, it wasn’t until the early 1960’s that webegan to see three influential forms of EC emerge:
– Evolutionary Programming (Lawrence Fogel, 1962),– Genetic Algorithms (Holland, 1962)– Evolution Strategies (Rechenberg, 1965 & Schwefel,
1968),
History
•The designers of each of the EC techniquessaw that their particular problems could besolved via simulated evolution.
• Fogel was concerned with solving prediction problems.
• Rechenberg & Schwefel were concerned with solving continousparameter optimization problems.
• Holland was concerned with developing robust adaptivesystems.
�Each of these researchers successfully developedappropriate ECs for their particular problemsindependently.
�In the US, Genetic Algorithms have become themost popular EC technique due to a book by David E.Goldberg (1989) entitled, “Genetic Algorithms inSearch, Optimization & Machine Learning”.
�This book explained the concept of Genetic Searchin such a way the a wide variety of engineers andscientist could understand and apply.
History•First Generation EC
• Evolutionary Programming (Fogel)• Genetic Algorithms (Holland)• Evolution Strategies (Rechenberg, Schwefel)
•Second Generation EC• Genetic Evolution of Data Structures (Michalewicz)• Genetic Evolution of Programs (Koza)• Hybrid Genetic Search (Davis)• Tabu Search (Glover)
History
Third Generation EC
• Artificial Immune Systems (Forrest)• Cultural Algorithms (Reynolds)• DNA Computing (Adleman)• Ant Colony Optimization (Dorigo)• Particle Swarm Optimization (Kennedy & Eberhart)• Memetic Algorithms• Estimation of Distribution Algorithms
History
• 1985: first international conference (ICGA)
• 1990: first international conference in Europe (PPSN)
• 1993: first scientific EC journal (MIT Press)
• 1997: launch of European EC Research Network EvoNet
EC in the early 21st Century
• 3 major EC conferences, about 10 small related ones
• 3 scientific core EC journals
• 750-1000 papers published in 2003 (estimate)
• EvoNet has over 150 member institutes
• numerous applications
• numerous consultancy and R&D firms
ApplicationsEvolutionary Computation has been successfullyapplied to a wide range of problems including:
• Aircraft Design,• Routing in Communications Networks,• Tracking Game Playing (Checkers [Fogel])• Robotics,• Air Traffic Control,• Design,• Scheduling,• Machine Learning,• Pattern Recognition,• Job Shop Scheduling,• VLSI Circuit Layout,• Design of Filters and Barriers,• Data-Mining,• User-Mining,• Resource Allocation,• Path Planning,
Darwinian Evolution: Survival of the fittest
• All environments have finite resources– (i.e., can only support a limited number of individuals)
• Life forms have basic instinct/ lifecycles geared towardsreproduction
• Therefore some kind of selection is inevitable
• Those individuals that compete for the resources mosteffectively have increased chance of reproduction
Fitness = HeightSurvival of the fittest
Darwinian Evolution : Diversity drives change
Phenotypic traits:– Behavior / physical differences that affect response to
environment– Partly determined by inheritance, partly by factors during
development– Unique to each individual, partly as a result of random
changesIf phenotypic traits:
– Lead to higher chances of reproduction– Can be inheritedthen they will tend to increase in subsequentgenerations,
leading to new combinations of traits N
Darwinian Evolution:Summary
• Population consists of diverse set of individuals• Combinations of traits that are better adapted tend
to increase representation in population• Individuals are “units of selection”
• Variations occur through random changes yieldingconstant source of diversity, coupled with selectionmeans that:
• Population is the “unit of evolution”• Note the absence of “guiding force”
Adaptive landscape metaphor (Wright, 1932)
• Can envisage population with n traits as existing ina n+1-dimensional space (landscape) with heightcorresponding to fitness
• Each different individual (phenotype) represents asingle point on the landscape
• Population is therefore a “cloud” of points, movingon the landscape over time as it evolves -adaptation
Example with two traits
Adaptive landscape metaphor
•Selection “pushes” population up the landscape
•Genetic drift:• random variations in feature distribution
(+ or -) arising from sampling error• can cause the population “melt down” hills, thuscrossing valleys and leaving local optima (oralternative global optima!)
Natural Genetics
• The information required to build a living organismis coded in the DNA of that organism
• Genotype (DNA inside) determines phenotype• [Genes � phenotypic traits] is a complex
mapping– One gene may affect many traits (pleiotropy)– Many genes may affect one trait (polygeny)
• Causality: Small changes in the genotype lead tosmall changes in the organism (e.g., height, haircolor)
• The effect of one gene on phenotype depends onthe values of other genes (opposite is orthogonality)
Genes and the Genome• Genes are encoded in strands of DNA called
chromosomes• In most cells, there are two (homologous) copies of
each chromosome (diploidy)• The complete genetic material in an individual’s
genotype is called the Genome• Within a species, most of the genetic material is the
same
Example: Homo SapiensHuman DNA is organized into chromosomesMost human body cells contain 23 pairs of
chromosomes which together define the physicalattributes of the individual:
Reproductive Cells
Gametes (sperm and egg cells) contain 23 individualchromosomes rather than 23 pairs
Cells with only one copy of each chromosome arecalled Haploid
Gametes are formed by a special form of cell splittingcalled meiosis
During meiosis the pairs of chromosomes undergo anoperation called crossing-over
Crossing-over during meiosis
Chromosome pairs align and duplicateInner pairs link at a centromere and swap parts of
themselves
� Outcome is one copy of maternal/paternalchromosome plus two entirely new combinations� After crossing-over one of each pair goes into eachgamete
Fertilization
Sperm cell from Father Egg cell from Mother
New person cell (zygote)
After fertilization
• New zygote rapidly divides creating many cells allwith the same genetic contents
• Although all cells contain the same genes,depending on, for example where they are in theorganism, they will behave differently
• This process of differential behavior duringdevelopment is called ontogenesis
MutationOccasionally some of the genetic material changes
very slightly during this process (replication error)This means that the child might have genetic material
information not inherited from either parentThis can be
– catastrophic: offspring in not viable (most likely)– neutral: new feature does not influence fitness– advantageous: strong new feature occurs
Redundancy in the genetic code forms a good way oferror prevention
Motivations for EC: 1Nature has always served as a source of
inspiration for engineers and scientistsThe best problem solver known in nature is:
– the (human) brain that created “the wheel,wars and so on”
– the evolution mechanism that created thehuman brain (after Darwin’s Origin ofSpecies)
Answer 1 � neurocomputingAnswer 2 � evolutionary computing
Motivations for EC: 2
Developing, analyzing, applying problem solvingmethods is a central theme in mathematics andcomputer science
Time for thorough problem analysis decreases
Complexity of problems to be solved increases
Consequence:Robust problem solving technology needed
Problem type 1 : Optimization
We have a model of our system and seek inputs thatgive us a specified goal
� e.g.– time tables for university, or hospital– design specifications, etc.
Optimization example 1: University timetabling
Enormously big search space
Timetables must be good
“Good” is defined by a number of competing criteria
Timetables must be feasible
Optimization example 2: Satellite structure
Optimized satellite designs forNASA to maximize vibrationisolation
Evolving: design structures
Fitness: vibration resistance
Problem types 2: ModelingWe have corresponding sets of inputs & outputs and
seek a model that delivers the correct output forevery known input
• Evolutionary machine learning
Modelling example: loan applicant creditibility
British bank evolvedcreditability model to predictloan paying behavior of newapplicants
Evolving: prediction models
Fitness: model accuracy onhistorical data
Problem type 3: Simulation
We have a given model and wish to know the outputsthat arise under different input conditions
� Often used to answer “what-if” questions in evolvingdynamic environments� e.g. Evolutionary economics, Artificial Life
Simulation example: evolving artificial societies
Simulating trade, economiccompetition, etc. to calibratemodels
Use models to optimizestrategies and policies
Evolutionary economy
EC in General
Pseudo Code
Types of EC