# Genetic Algorithms, Search Algorithms

Τεχνίτη Νοημοσύνη και Ρομποτική

23 Οκτ 2013 (πριν από 4 χρόνια και 6 μήνες)

80 εμφανίσεις

Genetic Algorithms, Search Algorithms

Jae C. Oh

Overview

Search Algorithms

Learning Algorithms

GA

Example

Brief History

Evolutionary Programming

Fogel in 1960s

Individuals are encoded to be finite state
machines

Intellgent Behavior

Evolutionary Strategies

Rechenberg, Schwefel in 1960s

Real
-
valued parameter optmization

Genetic Algorithms

Holland in 1960s

Crossover Operators

Current Status

Wide variety of evolutionary algorithms

No one seriously tries to distinguish them
except for some cases and purposes.

We will call all Evolutionary Algorithms

And I will call them Genetic Algorithms or
Evolutionary Algorithms for generic terms

Search

Search

Search

Notion of Search Space

Real world problem

Search space

Abstraction
-
> State Space

Exploring the state space for given
problem

Search Algorithms

The Peak

Search Space

Learning Algorithms

Finding (through search) a suitable
program, algorithm, function for a given
problem

Learning Algorithm

Training Data

(Experience)

Program

Learning Algorithms (function
Optimizations)

Problem instance

Set of Hypothesis

Hypothesis Space

Program Space

Function Space

The One??

Learning Algorithms (Digression)

How do we know the found hypothesis,
program, function, etc. are
the one

we
are looking for?

We don’t know for sure

Is there any mathematical way of telling
how good hypothesis is?

I.e., |h(x)

f(x)| = ?

Computational Learning Theory can tell us
this

Valiant (1984)

What are Genetic Algorithms?

Find solutions for a problem with the idea of
evolution.
Search

and
optimization

techniques
based on Darwin’s Principle of
Natural
Selection
.

Randomized
search and optimization algorithms

guided by the principle of Darwin’s natural selection:
Survival of fittest.

Evolve potential solutions

Step
-
wise refinement?

Mutations? Randomized, parallel search

Models natural selection

Population based

Uses fitness to guide search

Evolution is a search process

From the Tree of the Life Website,

University of Arizona

Orangutan

Gorilla

Chimpanzee

Human

Evolution is parallel search

AAGACTT

TG
GACTT

AAG
G
C
C
T

A
G
GGC
A
T

T
AG
C
CCT

A
G
C
ACTT

TAGCCC
A

TAG
A
C
T
T

AGC
G
CTT

AGCAC
AA

AGGGCAT

AGGACTA

T
GGACTA

AG
T
GAC
C
A

AG
G
G
CA
A

CAGCA
CCA

A
G
C
ACTA

TC
GCCCA

T
AG
GC
C
T
A

AG
TG
CTA

AG
T
AC
A
A

A
A
GGCAA

Genetic Algorithm Overview

1.
Starting with a subset of
n

randomly chosen
solutions ( )from the search space (i.e.
chromosomes). This is the
population

2.
This population is used to produce a next
generation

of individuals by reproduction

3.
Individuals with a higher
fitness

(|
-

|)have
more chance to reproduce (i.e. natural selection)

GA in Pseudo code

0 START

: Create random population of
n

chromosomes

1 FITNESS :
Evaluate fitness
f(x)

of each chromosome in

the population

2 NEW POPULATION

0 SELECTION :
Based on
f(x)

1 RECOMBINATION

:
Cross
-
over chromosomes

2 MUTATION :
Mutate

chromosomes

3 ACCEPTATION :
Reject or accept new one

3 REPLACE :
Replace old with new population: the new

generation

4 TEST :
Test problem criterium

5 LOOP :
Continue step 1

4 until criterium is

satisfied

GA vs. Specialized Alg.

Genetic Algorithms (GAs)

Problems

P

Specialized Algo.

GA

Specialized algorithms

best performance for special problems

Genetic algorithms

good performance over a wide range of problems

Randomized Algorithms

Guided random search technique

Uses the payoff function to guide
search

Global optima

local

optima

Hill Climbing

Evolutionary Algorithms?

Search Algorithms?

Learning Algorithms?

Function Optimization Algorithms?

They are fundamentally the same!!

Things needed for GAs

How do we represent individuals? Domain
Dependent

How do we interpret individuals?

Domain Dependent

What is the fitness function?

Domain Dependent

How are individual chosen for reproduction?

Choose better individuals (probabilistic)

How do individuals reproduce?

Crossover, Mutation, etc.

How is the next generation generated?

Replace badly performing individuals

Encoding Methods

11111110000000011111

Chromosome

B

10110010110011100101

Chromosome A

8

5

6

7

2

3

1

4

9

Chromosome B

1

5

3

2

6

4

7

9

8

Chromosome A

(left), (back), (left), (right), (forward)

Chromosome

1.235

5.323

0.454

2.321

2.454

Chromosome

Binary Encoding/Ternary Encoding

Permutation Encoding (TSP)

Real numbers, etc. Specialized

Fitness Function

A fitness function quantifies the optimality of a
solution (chromosome) so that that particular
solution may be ranked against all the other
solutions.

A fitness value is assigned to each solution depending
on how close it actually is to solving the problem.

Ideal fitness function correlates closely to goal +
quickly computable.

Example. In TSP, f(x) is sum of distances between
the cities in solution. The lesser the value, the fitter
the solution is

Producing Offspring

The process that determines which
solutions are to be preserved and allowed
to reproduce and which ones deserve to
die out.

The primary objective of the recombination
operator is to
emphasize the good solutions

and
eliminate the bad solutions

in a population,
while
keeping the population size constant.

“Selects The Best, Discards The Rest”.

Roulette Wheel Selection

1

2

3

4

Spin

Chromosome #

Fitness

1

15.3089

2

15.4091

3

4.8363

4

12.3975

Strings that are fitter are assigned a larger slot and hence
have a better chance of appearing in the new population.

GA in Action for 8
-
Queen

Fitness for 8
-
Queen?

Minimum conflict fitness function.

Theory (Schema Theorem)

Schema

Substring where some positions left
undecided

246*****

Instance of this schema: 24613587

Theorem: if the average of the instances
the schema is above the mean fitness of
the population, the number of instances of
the schema will increase over time.

Applications

Many many…

VLSI, TSP, Function Optimization, Data
mining, security, etc.