Parallel Genetic Programming

Francisco Ferná
Fernández de Vega
Grupo de Evolució
Evolución Artificial
University of Extremadura (Spain)

http://gea.unex.es

Summary:

Introduction.

History.

Parallel GP.

The Island Model.

Successful Applications.

Future.

1
Summary:

Introduction.

History.

Parallel GP.

The Island Model.

Successful Applications.

Future.

Some considerations

Genetic Programming can be considered a Machine
Learning system:
 “[machine learning] is the study of computer algorithms that
improve automatically through experience [Mitchell, 1996].

The emphasis is on learning (instead of on
knowledge).

The dream of computers that program themselves
[Samuel, 1963] could be reached soon.

Introduction 4

2
 What’s Genetic Programming?

According to Banzhaf et al*, GP is a system

that induce computer programs by
evolutionary means.

GP (Koza, 1992) is a kind of Evolutionary
Algorithm (EA). (Genetic Alorithms, Genetic Programming,
Evolutinary Strategies, Evolutionary Programming).

EAs can be seen as search techniques

(stochastic search technique).

*Banzhaf, W., Nordin, P., Keller, R.E., Francone, F.D. Genetic Programming, an
Introduction. Morgan Kaufmann 1998. Introduction 5

Different Search Techniques

Search Techniques

Enumeratives Calculus Stochastic

Based

Depth Breadth Hill Climbing Evolutionary

First first Algorithms

Simulated Neural Beam Genetic Genetic

Annealing Networks Search Programming Algorithms

As classified by Banzhaf et al
Introduction 6

3
How does an EA work

A summary:
 T=0;
 Initialize and evaluate [P(t)]
 While not stop_condition do

P´(t)=variation [P(t)]

Evaluate P´(t)

P(t+1)=select [P´(t),P(t)]

T=t+1
 end while

Introduction 7

How does GP work?

Individuals Population and

compete for Individuals
resources

Different
Reproduction
characters:
& Heredity

Genetic programming 8

4
How does GP work?

Genetic Operators:
•Crossover.
Functions
+ •Mutation.
* b
•Selection.
a -
•Reproduction.
b c

Terminals
GP Individual Fitness Function

Genetic programming 9

Genetic Programming

Before the operation

Crossover
point
+ +

* b a /

a - Subtrees to be b -
exchanged
b c b c

C
R After the operation

+
O +

* b a -
S a / b c

S b -

O b c

V Relies on the Building block hypothesis

Genetic programming 10
E

5
 Genetic Programming

Before the operation After the operation

+
+

* b
* b

a /
a -

M Mutation
Point b c b -

U + c

Subtree to
T be deleted
a -

b c
A
T
New subtree
I ramdomly
generated

O
Genetic programming 11
N

Genetic Programming

Fitness
Function
S
Fitness
E
values
L
Selection
E
C
T
I
Genetic programming 12
O

6
 Genetic Programming

f(x)

b(x) y(x)
…
g(x)

Genetic programming 13

Summary:

Introduction.

History.

Parallel GP.

The Island Model.

Successful Applications.

Future.

7
History

There are two main ideas behind Parallel
Eas:
 Increase performances:

in principle, by adding processors, memory and
interconnection networks and putting them to work
together on a given problem.
 Modifying the underlying algorithm can also help
in the finding of solutions.

Parallel EAs: History 15

History

Ideas involving both EAs and Parallel

Computing can be traced back to Holland,
1976.

But the field had to wait until early 1980s
when parallel implementations appear.

Grefenstete, 1981, was one of the first in
examaining some issues concerning parallel
implementations of Gas.

Parallel EAs: History 16

8
History

Other researchers began more systematic
studies: Gross, Cohoon, Tanese, Pettey,
Georges-Schleter, Mühlenbein, and
Manderick*.

They studied Hypercubes parallel
architecture, distributed models, theoretic
models, island models, and cellular model.

*See “Parallelism and Evolutionary Algorithms”, E. Alba & M. Tomassini, IEEE

Transactions on Evolutionary Computation, Vol. 6, NO. 5, Oct 2002, 443-462
for a review summary. Parallel EAs: History 17

Parallelism & EAs

Flynn model is still widely accepted for classifying
computer architectures.

The taxonmy is based on the notion of instruction
and data stream:
 SISD: Single instruction, Single Data stream.
 SIMD: Single Instruction, Multiple Data stream (the
preferred model).
 MISD: Multiple instruction, single data stream.
 MIMD: Multiple instruction, multiple data stream.

Shared or Distributed Memory.

Parallel GP 18

9
Taxonomy

Parallel EAs can be classyfied attending to different features.

M Nowostawski and R. Poli 1999:
 Master/Worker: A single population and the fitness evaluation of
multiple individuals in parallel.
 Static subpopulations with migration.
 Static overlapping subpopulations without migration.
 Massively Parallel genetic algorithms.
 Dynamic demes.
 Parallel Steady-state Ga
 Parallel Messy Ga
 Hybrid methods.

Parallel EAs: History 19

Taxonomy

M. Tomassini, 1999:
 Global parallel evolutionary algorithm
(parallelization at the fitness level) also called by
the author master/worker model (coarse-grained
model).
 Island distributed evolutionary algorithms
(population based approach).
 Cellular evolutionary algorithm (fine-grained
model).

Parallel EAs: History 20

10
Structured EAs

Structured populations has been used for improving
EAs.

Two main types of algorithms:
 Distributed EAs.
 Cellular EAs.

On the oposite side, panmictic EA is the classic
model.

None of the models require a parallel
implementation.

Parallel EAs: History 21

Structured EAs

Distributed EAs Cellular EAs

Subpopulation Individuals

Migrating Interactions
individuals among individuals

Parallel EAs: History 22

11
Nonstandard Structured EAs

We could use different
parameters/representations in different
subpopulations (Tanese, Lin & Punch & Goodman, Herrera &
Lozano & Moraga).

These algorithms are sometimes called

heterogenous.

Parallel EAs: History 23

Summary:

Introduction.

History.

Parallel GP.

The Island Model.

Successful Applications.

Future.

12
 Parallel Models

Parallelising at the Why should we use a parallel

fitness level. model?
•We want to increase
performances.

Master
Process

How could we parallelise?

Processor Processor Processor Processor • At the individual level.
#1 #2 #3 … #n

• At the population level.

This model is also called •At the fitness evaluation
“global model”. level

Parallel GP 25

Parallel Models
Individuals

Parallelising at the Relationships

population level (also

called island model or
coarse-grain model).

Subpopulation

Migrating
individuals
Fine-grained model (also
called Grid or Cellular
model)

Parallel GP 26

13
Parallel GP: A review

Juillé and Pollack 1995 presented one of the first
attempts to parallelize GP.

They implemented a global model, although also
presented some result using sub-populations.

The proposal was somehow specific for the SIMD
model they were using.

One of their aims was to reduce interprocessor
communication.

Parallel GP 27

Parallel GP : A review

P. Tuffs, 1995, presented the same year

another master/worker parallel version of GP.

He approached a classification problem by
means of GP. -the development of a system to do data mining on a
fairly large (multi-gigabyte) database of credit-card transactions. The task was to
classify customers and predict their future behavior-

Probably it was the first attempt to parallelize

GP (the book correspond to year 1993, although was published on
1995).

Parallel GP 28

14
Parallel GP : A review

Andre and Koza 1996 presented another Parallel
GP Implementation using a network of 66
transputers (VLSI device containing 32 bit on-chip processor, memory and links)

They employed the island-model.

An appropriate migration rate showed improvement
in the computational effort required for the Boolean
5-parity function.(64 demes, SubPop_size=500, Migration_rate=5%)

The main conclusion was that Parallel GP could
achieve super-linear speedups.

Parallel GP 29

Parallel GP : A review

K. Stoffel and L. Spector 1996, implemented a
parallel version of GP using linear programs
evaluated by means of a stack.

Several processors generate independently its own
segment of the next generation.

They implemented crossover in parallel: Individuals
from different processors may undergo crossover.

Parallel GP 30

15
Parallel GP : A review

Oussaidène et al 1997, presented a parallel
implementation of GP for trading model induction.

They employed the global model architecture
(parallelization at the fitness level), employing a
master/worker model, where each node from the
network is in charge of evaluating individuals comming
from a master node.

The master node is in charge of the main GP
algorithm.

The model may undergo a load imbalance problem.

Parallel GP 31

Parallel GP: A review

Results offered by Koza were questioned later by W.

F. Punch 1998. His experiments on the Royal Tree
problem were not so optimistics. His main
conclusions were that multiple-solution problems
would be more amenable to multiple populations
than single-solution problems.

On the other hand, non-deceptive problems would
be more amenable to multiple populations than
deceptive problems.

He only tried a set of parameters for the parallel
model.

Parallel GP 32

16
Parallel GP : A review

Fernandez et al 1999, presented
experimental results on an island-model
implementation of GP.

Although results were preliminar, this is the
first time some important parameters of the
island-model are tested (communication
topology).

Parallel GP 33

Parallel GP : A review

Folino et al, 2000, presented a new
implementation of GP using the cellular
model.

Fernández et al, 2000, studied more deeply
the relationship among several important
parameters for the island-model (subpop
size, number of subpop, communication
topology).

Parallel GP 34

17
Parallel GP : A review

The latest results on both the island and

cellular GP models have been presented very
recently:
 F. Fernández et al., 2003, described latest results
with the island-model, while G. Folino et al, 2003,
presented a Cellular Scalable implementation for
GP. Their results are compared with previously
described results using parallel GP.

Parallel GP 35

Parallel GP tools

Many GP tools allow the use of “demes” but

simmulated in a sequential fashion.

There have been several parallel
implementations during the last few years.

Several languages (C, java, C++) and
communication frameworks (sockets, java
rpc, pvm, mpi …) have been employed.

Parallel GP 36

18
Communication Tools

PVM: Parallel Virtual Machine (V.S. Sunderam,

“PVM: A framework for parallel distributed computing, “ J. Concurr. Practice and
Experience, vol.2, no.4, pp. 315-339, 1990”)

MPI: Message Passing Interface.(Message

Passing Interface Forum, “MPI: A message-passing interface standard,” Int. J.
Supercomput, Applic., vol.8, no.3-4, pp.165-414, 1994)

GLOBUS.

Others (Sockets, Java-RMI…)

Parallel GP 37

Parallel GP tools

Chong, 1998, presented DGP, a java based distributed approach

to genetic programming on the Internet.

F. Fernández et al, 1999, developed a parallel GP tool
implementing the island-model, and communicating
subpopulations by means of PVM. This tool was later improved
by means of MPI (Fernández et al, 2000).

Spezzano et al, 2001, presented CAGE: A tool for parallel
genetic programming applications. They implemented the
cellular model.

Classic LilGP software (Punch) also has a couple of parallel
version implemented using PVM and MPI (see Fernández
parallelilgp)

Parallel GP 38

19
Parallel GP tools

DREAM project: It is aimed at providing a framework
for evolutionary computation.

It allows distributed computing.

Any Evolutionary Algorithm could be used, by
adjusting some parameters, within DREAM.

Founded by European Union.

See:
http://www.dcs.napier.ac.uk/~benp/dream/dream.ht
m

Parallel GP 39

Parallel GP tools

Paradiseo: Parallel and Distributed Evolving
Objects.

It is based on EO (Evolutionary Computation
Framework).

Includes tools for:
 Population Based Metaheuristics.
 Single Solution Based Metaheuristics.
 Multi-Objective Metaheuristics.
 Parallel and Distributed…

Parallel GP 40

20
Parallel GP tools

ECJ: A Java-based Evolutionary

Computation Research System.

Includes asynchronous Island Model over
TCP/IP.

Multiobjective Optimization.,

Summary:

Introduction.

History.

Parallel GP.

The Island Model.

Successful Applications.

Future.

21
Island Model
Important Parameters:

• Size of Subpop.

• Topology.

• Communication rate.

• Granularity.

• Synchronization.

Important concern:

• Comparing results.

The Island Model 43

Island Model – Comparing results

Traditionally experimental results are shown comparing
Fitness/Generation.

Two reasons for avoiding this kind of comparisons:
 The Bloat phenomenon in GP.

 Populations with different size

Generations
require different time to
evaluate a generation.

The Island Model 44

22
Island Model – Comparing Results

The Island Model 45

Island Model – Measuring Results

Proposal: Evaluate Fitness/Effort (for

convergence) or Fitness/Time (for speedup)
(Fernández, Galeano, Gómez).

PEg=i*p*avg_lengthg

p:the number of populations

Computational i:the number of individuals per
population
effort: avg_lengthg: the average length of
individuals in all the populations in
The total number generation g.
of nodes GP has The computational effort Eg at a
generation g is:
evaluated for a
given number of Eg=PEg+PEg-1+ ... + PE1 + PE0
generations. The Island Model 46

23
Island Model – Comparing Results

Island Model - Topology

Andre and Koza 1996 connected each
subpopulations with 4 neighbors in the N, E,
W, S directions.

The Island Model 48

24
Island Model - Topology

Punch, 1998, used a typical Island model
with ring topology.

He employed that
topology for his
experiments but no
comparisons with
different topologies were
Ring Topology provided in the paper.

The Island Model 49

Island Model - Topology

Fernandez et al 2000 introduces a random topology and
compare it with grid and ring topology.

Ring Topology Random Topology: It changes dynamically.

The main conclusion is that if the remaining parameters

stay fix, there are no significant differences when
changing topology.
The Island Model 50

25
Island Model - Topology
Ant

80

70
Grid
60
Circle
50
Random
Grid 9-60
Fitness

40 Circle 9-60
Random 9-60

30

20

10

0
0 500000 1000000 1500000 2000000 2500000 3000000 3500000 4000000
Effort

The Island Model 51

Island Model - Topology

Evenp 5

14

12

Circle
10
Random
8
Grid
Fitness

Random 9-500
Circle 9-500
Grid 9-500
6

0
0 2000000 4000000 6000000 8000000 10000000 12000000
Effort

The Island Model 52

26
Island Model – Migration Rate

How many individuals should migrate each
migration step?

Deppending on the number of individuals, results
are different. The limits are:
 0 individuals migrating (isolated populations)
 All the individuals migrating.

Different migration rates applied in literature:
 Juillé & Pollack: 1 migrating individual per subpopulation.
 Andre and Koza: 0%-8% migrating individuals per subpop.
 Punch 1998: 2 individuals.

The Island Model 53

Island Model – Migration Rate

Fernandez et al 2003: best migration rate is

between 5% and 10% in 4 test problems (2
classic and 2 real-life problems):
 Even Parity 5.
 Ant Problem.
 Routing and Placing circuits on FPGAs.
 Medical Diagnosing.

The Island Model 54

27
Island Model – Migration Rate
5 Populations
100 Individuals/Pop

ANT Problem

The Island Model 55

Island Model – Migration Rate

5 Populations
100 Individuals/Pop

FPGA Problem

The Island Model 56

28
Island Model – Migration Frequency

Both Juillé & Pollack and also Andre and Koza

employ migration every generation.

In Punch 1998, subpop. wait for 10 generations
before the migration step.

In Fernandez et al 2003, a wider study have been
carry on, comparing different frequencies.

Best convergence results appear when about 10%
of individuals from each subpopulation are sent
every 5-10 generations.

The Island Model 57

Island Model – Migration frecuency

5 Populations
100 Individuals/Pop

Evenp-5 Problem

The Island Model 58

29
Island Model – Migration summary

5 Populations
100 Individuals/Pop

FPGA Problem

The Island Model 59

Island Model – Migration Rate

5 Populations
100 Individuals/Pop

ANT Problem

The Island Model 60

30
Island Model – Migration Summary

For large values of the grain, exchange

individuals less frequently.

For low values of the grain, exchange more
frequently.

Recommendation: Exchange 10% of the
population every 10 generations.

The Island Model 61

Island Model – Subpop. Size

Experiments presented by Andre & Koza,

make use of the large computational
capability of the network they employed.

32000 individuals are distributed among 64
demes, each with 500 individuals.

Punch 1998, employed 5 populations, 200
individuals each, and also 7 populations, 700
individuals each.

The Island Model 62

31
 Island Model – Subpop. Size

F. Fernández et al., 2003, presents a set of
trials.

Conclusions:
 There is a number of individuals with which best
results are obtained (regardless of the number of
subpops).
 We must carefully select the number of subpops,
not any number of populations obtain the same
results.

The Island Model 63

Island Model – Subpop. Size

(250-500)

EVENP 5
T The Island Model 64
64

32
Island Model – Subpop. Size

(250-500)

EVENP 5
The Island Model 65

Island Model – Synchronisation

Synchronism is an important issue when using

Parallel GP: different individuals may require
different processing time for their evaluation.

Two models:
 Synchronous: Exchange step takes place at a given
generation.
 Asynchronous: Populations send individuals when they are
ready, and check every generation if new incoming
individuals are awaiting.

The Island Model 66

33
Island Model – Synchronisation

Andre and Koza worked with an asynchronous

model: each generation is typically working on
different generations after a few ones.

Dracopoulos and Kent, employed the synchronous
model in both the global and island models.

Fernández et al, 2002, presented a study comparing
synchronous and asynchronous models in
monoprocesor systems.

The Island Model 67

Synchronous - Asynchronous

Results obtained using 1 processor

•Synchronous model better for monoprocessor system.

•Asynchronous model better for parallel systems.

The Island Model 68

34
Synchronous - Asynchronous

Tongchim and Chongstitvatana, presented a
comparison among models using a restricted
migration policy.

Their results only focus on a problem, and
shows better performance with the
asynchronous model.

The Island Model 69

Island Model - Bloat

The Island Model
seems to prevent the
bloat phenomenon.

The Island Model 70

35
Some comments on diversity

•Genotypic diversity decreases.

•Phenotypic diversity improves.

What about fault tolerance?

Fault tolerance is an important issue.

Different techniques have been employed:
 Check pointing.
 Redundancy.
 Others…

Is GP Fault tolerant?

36
What about fault tolerance?

The Cellular Model

Folino et al, 2003, have presented a

comparison with panmitic and island model
approach.

The method provides results of similar quality
than the island model (an small error in the comparisons seems
to favor the cellular model in the paper, but a detailed revision shows that
results are similar).

They apply the model to induce decision trees.

The Island Model 74

37
The Cellular Model

Each individual is
located in a grid
position.

Individuals interact only
with their neighboring
ones.

The Island Model 75

Summary:

Introduction.

History.

Parallel GP.

The Island Model.

Successful Applications.

Future.

38
Some Applications

C. Miccio et al, 1995, described an implementation

on a T3D computer for inducing binary decision
diagrams.

M. Oussaidène et al, presented an application to
trading model induction.

F. Fernández, 2001, described a proposal for
solving the problem of Placement and Routing of
circuits on FPGAs.

Folino apply the cellular model to generate decision
trees.

The Island Model 77

Some Applications

Koza et al 2000, presents a list of “Human-
Competitive results obtained by Means of Genetic
Programming”, including:
 Synthesis of Analog Circuits.
 Synthesis of PID controllers.
 Applications to biomedicine (protein detection).
 Previously patented inventions, reinvented.
 Some patented invention.

The Island Model 78

39
Summary:

Introduction.

History.

Parallel GP.

The Island Model.

Successful Applications.

Future.

Future

Some topics for future research:
 Theoretical models.
 Heterogeneous models.
 Improvements by means of better scheduling
policies.
 Bloat phenomenon.

Future 80

40
References

Tufts, P. “Parallel Case Evaluation of Genetic Programming. In
1993 Lectures in Complex Systems, Es. L. Nadel and D. Stein,
SFI Studies in the Science of Complexity, Lec. Vol. VI, Addison
Wesley, 1995, pp.591-596.

H. Juillé & J. B. Pollack, Parallel Genetic Programming and fine-
grained SIMD architecture” in Working Notes for the AAAI Symp.
Genetic Programming, E. V. siegel and J. R. Koza, Eds.
Cambridge, MA: mit, Nov. 10-12, 1995, AAAI, pp. 31-37.

Stoffel, K., and L. Spector. 1996. High-Performance, Parall,
Stack-Based Genetic Programming. In Koza, John R., Goldberg,
David E., Fogel, David B., and riolo, Rick L. (eds) Genetic
Programming 1996: Proceedings of the First Annual
Conference, 224-229. Cambridge, MA: The MIT Press.

References

D. Andre & J. R. Koza, “Parallel Genetic Programming: A

Scalable Implementation Using the Transputer Network
Architecture”, in P. Angeline and K. Kinnear editors, Advances in
Genetic Programming 2, Cambridge, MA, 1996. The MIT Press.

D. C. Dracopoulos and S. Kent, “Bulk Parallelisation of Genetic
Programming”, in Proceedings 3rd Int. Workshop Applied Parallel
Computing, pp.216-226. Springer Verlag 1996.

M. Oussaidène, B. Chopard, O. Pictet, and M. Tomassini,
“Parallel Genetic Programming and its application to trading
model induction”, Parallel Comput. Vol. 23, pp. 1183-1198, 1997.

41
References

P. Martin, “A Hardware Implementation of a Genetic

Programming System Using FPGAs and Handle-C”. In Genetic
Programming and Evolvable Machines, 2, 317-343, 2001.

M. Tomassini, “Parallel and Distributed Evolutionary Algorithms:
A Review”, in Engineering and Computer Science, J. Wiley and
Sons, Chichster, 1999, K. Miettinen et al (eds), pp. 113-133,
1999.

M. Tomassini & E. Alba, “Parallelism and Evolutionary
Algorithms”, in IEEE Transactions on Evolutionary Computation,
Vol. 6, No. 5, pp. 443-462.

References

R. Poli, “Evolution of graph-like programs with parallel distributed

genetic programming. In proceedings of the 7th International
Conference on Genetic Algorithms, T. Bäck (ed.), Morgan
Kaufmann, San Fco, CA, 1997, 346-353.

M. Nowostawski and R. Poli, “Parallel genetic algorithms
taxonomy”, in Proc. Third Int. Conf. Knowledge-Based Intell. Inf.
Engrg. Syst. KES’99, IEEE Computer Society: New York, Aug.
1999., pp.88-92.

G. Folino, C. Pizzuti, G. Apezzano, “A Scalable Cellular
Implementation of Parallel Genetic Programming”, in IEEE
Transactions on Evolutionary Computation, Vol. 7, No. 1, Feb
2003.

42
References

Heywood, M.I and Zincir-Heywood, A. N., “Register based
genetic programming on FPGA computing platforms”, in Proc
EuroGP’2000, R. Poli, W. Banzhaf, W. B. Langdon, J. F. Miller,
P. Nordin, and T. C. Fogarty (eds.), LNCS, vol. 1802, Edinburgh,
Springer: Berlin, 15-16 Apr., 2000, pp-44-59.

F. Fernández & M. Tomassini, “Solving the Ant and the Even
Parity-5 problems by means of parallel genetic programming “,
Late Breaking Papers at the 1999 Genetic and Evolutionary
Computation Conference, Brave and We (eds.), pp 88-92.

References

Folino et al, “Genetic Programming and Simulated Annealing: a
hybrid method to evolve dicision trees.” in R. Poli. Et al (eds),
Proceedings EuroGP 2000, LNCS 1802, pp. 294-303. Springer
Verlag, Berlin, 2000.

G. Spezzano, G. Folino, and C. Pizzuti, “CAGE: A tool for
parallel genetic programming applications”, in Proc.
EuroGP’2001 Genetic Programming, vol. 2038, J. Miller et al
eds., 2001, pp- 64-73.

F. Fernandez et al 2000 “Experimental study of Multipopulation
parallel genetic programming” in R. Poli. Et al (eds), Proceedings
EuroGP 2000, LNCS 1802, pp. 283-293. Springer Verlag, Berlin,
2000.

43
References

F. Fernández et al., 2003, “An Empirical Study of Multipopulation

Genetic Programming “, en Genetic Programming and Evolvable
Machines, Kluwer Accademic Press, 2003, vol. 4, no. 1, pp. 21-
51.

G. Folino et al, 2003, “A Scalable Cellular Implementation of
Parallel Genetic Programming” in IEEE Transactions on
Evolutionary Computation, vol. 7, no. 1, 2003, pp. 37-53.

F.S. Chong, “A Java based disributed approach to genetic
programming on the Internet”, Master’s thesis, Univ. Birmingham,
Birmingham, AL, 1998.

References

F. Fernández, J.M. Sánchez, M. Tomassini, and J. A. Gómez, “A

parallel genetic programming tool based on pvm”, in vol. 1697
LNCS pp. 241-293.

F. Fernández, M. Tomassini, L. Vanneschi and L. Bucher, “A
distributed computing environment for genetic programming
using MPI”, in vol 1908, LNCS, pp. 322-329.

W. Punch, lilgp: http://garage.cse.msu.edu/software/lil-gp/lilgp-
index.html

W. B. Langdon, Distributed Genetic Programming, available at
ftp://ftp.cs.bham.ac.uk/pub/authors/W.B.Langdon/gp-code/DGP/

44
References

W. F. Punch, 1998, “How Effective are Multiple Populations in

Genetic Programming”, Proceedings of the third Annual Genetic
Programming Conference, July 22-25 1998. pp. 313-318.

F. Fernández, G. Galeano, J. A. Gómez, “Comparing
Synchronous and Asynchronous Parallel and Distributed Genetic
Programming models”, in LNCS 2278, Pp. 326-335. Springer
Verlag 2002.

J. R. Koza, M. A. Keane, J. Yu, F. H. Bennet, W. Mydlowec,
“Automatic Creation of Human Competitive Programs and
Controllers by Means of Genetic Programming”. In Genetic
Programming and Evolvable Machines 1, 121-164(2000). Kluwer
Academic Publishers.

References

C. Miccio, E. Sánchez, M. Tomassini, “Parallel Genetic
Programming Induction of Binary Decision Diagrams”, in EPFL
Supercomputing Review, N. 7, 1995. Echle Politechnique
Federale de Lausanne, Switzerland.

F. Fernández, 2001, Fernández, "Distributed Genetic
Programming Models with application to Logic Synthesis on
FPGAs", PhD Thesis. University of Extremadura. February
2001.

F. Fernández, E. Cantú-Paz, Introduction Special Issue on
Parallel Bioinspired Algorithms. Journal of Parallel and
Distributed Computing, Elsevier Volume 66, Issue 8 , August
2006, Pages 989-990

45