Cultural Algorithm - Genetic Algorithms - Related Techniques
1. Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques
eng. D. M.
Condurachi
Cultural Algorithm
Background &
Definition
Genetic Algorithms. Related Techniques
Advantages &
Disadvantages
Communication
Protocol &
Algorithm eng. Daniel Mihai Condurachi
General Features
Suitable Problems
Department of Computer Science
Communication “AL.I.Cuza” University of Ia¸ i
s
Protocols
Ia¸ i, România
s
Embedding Other
Methods
Future Directions March 26, 2008
Appendices
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 1 / 17
2. Background
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques
eng. D. M.
Condurachi
Background &
Definition cultural evolution enables societies to evolve or adapt to their
Advantages & environments at higher rates than biological evolution based
Disadvantages
Communication
upon genetic inheritance alone
Protocol &
Algorithm over time humans have developed a unique set of capacities
General Features that support the formation, encoding, and transmission of
Suitable Problems cultural information [4]
Communication
Protocols
Embedding Other
Methods
Future Directions
Appendices
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 2 / 17
3. Definitions
Cultural
Algorithm -
Genetic
Algorithms.
Definition
Related
Techniques Culture is a system of symbolically encoded conceptual
eng. D. M.
Condurachi
phenomenon that are socially and historically transmitted within
and between populations [3].
Background &
Definition
Advantages &
Disadvantages
Definition
Communication
Protocol &
Cultural algorithms are computational modelsa of cultural evolution
Algorithm [4].
General Features
a modele de calcul
Suitable Problems
Communication
Protocols
Embedding Other Definition
Methods
Future Directions
Cultural algorithms are evolutionary algorithms that support an
Appendices additional mechanism for information extraction during the
execution of the algorithm, avoiding the need to encode the
information a priori [1].
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 3 / 17
4. Advantages
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques
eng. D. M.
Condurachi it reduces the need for immature individuals to waste energy
Background & by bypassing trial and error iterations usually required to
Definition
acquire information about the environment
Advantages &
Disadvantages it enables the transmission of more information than any
Communication
Protocol &
individual genome may feasibly contain
Algorithm
culture affords populations with
General Features
Suitable Problems
flexibility - cultural information can be transmitted faster than
Communication
genetic material
Protocols stability - culture is capable of persisting beyond the lifetime of a
Embedding Other
Methods
single individual [2]
Future Directions
Appendices
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 4 / 17
5. Communication Protocol
Cultural
Algorithm - interface between the population and belief space
Genetic
Algorithms. best individuals of the population can update the belief space
Related
Techniques via the update function
eng. D. M. the knowledge categories of the belief space can affect the
Condurachi
population component via influence function. The influence
Background &
Definition function can affect population by altering the genome or the
Advantages & actions of the individuals
Disadvantages
Communication
Protocol &
Algorithm
General Features
Suitable Problems
Communication
Protocols
Embedding Other
Methods
Future Directions
Appendices
Figure: Population and Belief Space and the Connection Between the two
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 5 / 17
6. Evolutionary perspective
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques Modeling the cultural evolution process:
eng. D. M.
Condurachi micro-evolutionary perspective - transmission of behaviors or
Background & traits1 between individuals in a population
Definition
Advantages &
macro-evolutionary perspective (formation of generalized
Disadvantages beliefs based upon individual experiences)
Communication
Protocol & Each individual can be described:
Algorithm
General Features a set of traits or behaviors and a mappa (symbolization of an
Suitable Problems individuals past experience and forecasts concerning future
Communication experience)
Protocols
Embedding Other or generalized description of their experiences
Methods
Future Directions
The same happens to the whole society
Appendices
1 trasaturi
˘ ˘
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 6 / 17
7. Group mappa
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques
eng. D. M.
Condurachi
Background & individual mappa can be
Definition
Advantages &
merged,
Disadvantages generalized,
Communication
Protocol &
specialized
Algorithm
General Features
in order to form group mappa
Suitable Problems group mappa serves to direct the future actions of the group
Communication and its individuals
Protocols
Embedding Other
Methods
Future Directions
Appendices
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 7 / 17
8. Basic Algorithm
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques
Begin
eng. D. M.
Condurachi t = 0;
Background &
initialize Population Space POP(t);
Definition
initialize Belief Space BLF(t);
Advantages &
Disadvantages repeat until termination condition achieved;
Communication perform actions of the individuals in POP(t);
Protocol &
Algorithm evaluate each individual by using the fitness function;
General Features select the best individuals to become parents;
Suitable Problems create new generation of offspring by mutation & crossover;
Communication
Protocols
delele not so fit members to make room for the new ones;
Embedding Other
bLF(t) alters the genome of the offspring - influence function;
Methods
best individuals can update the BLF(t) - acceptance function;
Future Directions
End.
Appendices
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 8 / 17
9. General Features
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques
dual Inheritance (at population and knowledge levels)
eng. D. M.
Condurachi
knowledge guide evolution of the population
Background &
supports hierarchical structuring of population and belief
Definition spaces
Advantages &
Disadvantages domain knowledge separated from individuals
Communication
Protocol & supports self adaptation at various levels
Algorithm
General Features
evolution can take place at different rates at different levels
Suitable Problems (“Culture evolves 10 times faster than the biological
Communication component”)
Protocols
Embedding Other
supports hybrid approaches to problem solving
Methods
Future Directions
a computational framework2 within which many all of the
Appendices
different models of cultural change can be expressed
2 cadru
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 9 / 17
10. General Features
Cultural
Algorithm -
Genetic
Algorithms. can support the emergence of hierarchical structures in both
Related
Techniques the belief and population spaces
eng. D. M.
Condurachi
Background &
Definition
Advantages &
Disadvantages
Communication
Protocol &
Algorithm
General Features
Suitable Problems
Communication
Protocols
Embedding Other
Methods
Future Directions
Appendices
Figure: Hierarchical structures in both the belief and population spaces
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 10 / 17
11. Suitable Problems
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques significant amount of domain knowledge (e.g. constrained
eng. D. M. optimization problems)
Condurachi
complex Systems where adaptation can take place at various
Background &
Definition levels, at various rates in the population and belief space
Advantages &
Disadvantages knowledge is in different forms and needs to be reasoned
Communication about in different ways
Protocol &
Algorithm hybrid systems that require a combination of search and
General Features
knowledge based frameworks
Suitable Problems
Communication
problem solution requires multiple populations and multiple
Protocols belief spaces and their interaction
Embedding Other
Methods hierarchically structured problem environments where
Future Directions hierarchically structured population and knowledge elements
Appendices
can emerge
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 11 / 17
12. Acceptance Function
Cultural
Algorithm -
Genetic
Algorithms. all individuals are used to update constraints knowledge3
Related
Techniques the top 20% (eminent individuals) are use are used ot update
eng. D. M.
Condurachi
the normative knowledge
Background & Future development:
Definition
Advantages &
use interference rules to adjust the classification of active
Disadvantages cells:
Communication
Protocol &
1 feasible
Algorithm 2 infeasible
General Features 3 semi-feasible
Suitable Problems
Communication
adjust hierarchical structure based upon the above
Protocols
split semi-feasible cell into smaller cells when the number of
Embedding Other
Methods individuals becomes too high
Future Directions
merge, recombine infeasible children into the original parent;
Appendices
then decompose the parent in a different way obtaining
semi-feasible children
3 cuno¸ tin¸ele
s t de constrângere
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 12 / 17
13. Influence Function
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques
eng. D. M. guides the migration of individuals from less productive cells
Condurachi
(infeasible), to ones that are more productive (semi-feasible
Background &
Definition and feasible)
Advantages &
Disadvantages
eminent cells = semi-feasible and feasible cells with eminent
Communication
individuals
Protocol &
Algorithm highlights the migration from ordinary cells to eminent cells
General Features
Suitable Problems
HOW?
Communication 1 perturb individuals a little in eminent cells
Protocols
Embedding Other
2 move individuals in infeasible cells to feasible ones
Methods
Future Directions
3 move individuals from ordinary to eminent cells
Appendices
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 13 / 17
14. Population Models Used
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques
eng. D. M.
Condurachi
Background &
Genetic Algorithms (Concept learning, optimization)
Definition
Genetic Programming (Evolving agent strategies)
Advantages &
Disadvantages
Evolutionary Programming (Real valued function optimization)
Communication
Protocol & Evolution Strategies (Robot soccer plays)
Algorithm
General Features Memetic models (Evolution of agriculture)
Suitable Problems
Agent based modeling (Evolution of the state, Environmental
Communication
Protocols Impact)
Embedding Other
Methods
Future Directions
Appendices
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 14 / 17
15. Knowledge Models Used
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques
eng. D. M. Schemata
Condurachi
Binary valued (Maleticconcept learning, Boole problem, data
Background &
Definition mining)
Advantages & Real-valued interval schemata (Chang:unconstrained
Disadvantages
optimization)
Communication
Protocol &
Fuzzy Real-valued schemata
Algorithm Regional Schemata ((Xidong Jin)constrained optimization)
General Features
Suitable Problems
Semantic Networks (DLMS:Rychtyckyj)
Communication Graphical Models (GP:Zannoni, Ostrowski)
Protocols
Embedding Other
Logical and Rule Based models (HYBAL(Sverdlik), Fraud
Methods
Detection (Sternberg), Lazar (Data mining))
Future Directions
Appendices
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 15 / 17
16. Variations
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques
eng. D. M.
Condurachi
Background &
Definition Space Guided Genetic Algorithms (VGA)
Advantages &
Disadvantages the micro-evolutionary process is modeled using genetic
Communication algorithms
Protocol &
Algorithm
the belief space represents schemata or generalizations of
General Features
the individual chromosomes based upon their behaviors
Suitable Problems
Communication
Protocols
Embedding Other
Methods
Future Directions
Appendices
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 16 / 17
17. Future Directions
Cultural
Algorithm -
Genetic
Algorithms.
Related
Techniques
eng. D. M.
Condurachi
Background &
Definition
Integrating Multiple Representations and Population Models
Advantages & Parallelization
Disadvantages
Communication Belief Space Evolution
Protocol &
Algorithm Designing Cultural Systems
General Features
How does a Culture’s structure and content reflect its problem
Suitable Problems
Communication
solving environment (Saleem)
Protocols
Embedding Other
Methods
Future Directions
Appendices
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 17 / 17
18. Cultural
Algorithm - Ricardo Landa Becerra.
Genetic
Algorithms. Use of Domain Information to Improve the Performance of an
Related
Techniques Evolutionary Algorithm.
eng. D. M. PhD thesis, Center for Research and Advanced Studies of the
Condurachi
Nationl Polytechnic Institure of Mexico, Computer Science
Background &
Definition
Department, 2007.
Advantages &
Disadvantages Dara Curran.
Communication An Empirical Analysis of Cultural Learning: Examining fitness,
Protocol &
Algorithm diversity and changing environments in populations of
General Features game-playing neural network agents.
Suitable Problems PhD thesis, The Department of Information Technology of the
Communication
Protocols
National University of Ireland, Galway, 2006.
Embedding Other
Methods W. Durham.
Future Directions Co-Evolution: Genes, Culture, and Human Diversity.
Appendices Stanford University Press, 1994.
Robert G. Reynolds.
An introduction to cultural algorithms.
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 17 / 17
19. Cultural Technical report, Department of Computer Science 431 State
Algorithm -
Genetic Hall Wayne State University, 1994.
Algorithms.
Related http://ftp.cerias.purdue.edu/pub/doc/EC/EA/
Techniques
papers/cult94.ps.gz
eng. D. M.
Condurachi email: reynolds@cs.wayne.edu.
Background &
Definition
Advantages &
Disadvantages
Communication
Protocol &
Algorithm
General Features
Suitable Problems
Communication
Protocols
Embedding Other
Methods
Future Directions
Appendices
eng. D. M. Condurachi (UAIC) Cultural Algorithm - Genetic Algorithms. Related Techniques March 26, 2008 17 / 17