## An Analysis of Island Models in Evolutionary Computation (2005)

Venue: | In Proceeding of the 2005 workshops on Genetic and evolutionary computation |

Citations: | 5 - 0 self |

### BibTeX

@INPROCEEDINGS{Skolicki05ananalysis,

author = {Zbigniew Skolicki},

title = {An Analysis of Island Models in Evolutionary Computation},

booktitle = {In Proceeding of the 2005 workshops on Genetic and evolutionary computation},

year = {2005},

pages = {26}

}

### OpenURL

### Abstract

A need for solving more and more complex problems drives the Evolutionary Computation community towards advanced models of Evolutionary Algorithms. One such model is the island model which, although the subject of a variety of studies, still needs additional fundamental research. In my Ph.D. thesis I am aiming at studying the behavior of island models with regard to the amount of cooperation between islands, the level of heterogeneity and the difficulty of the problem being solved. This paper presents the main ideas and gathers preliminary results.

### Citations

222 | Efficient and Accurate Parallel Genetic Algorithms
- Cantu-Paz
- 2000
(Show Context)
Citation Context ...(among others the population size, the connectivity, the migration rate, the takeover time and the selection intensity), some of them on more theoreticalmodels and others in more experimental setups =-=[4]-=-. There exist models where migrations are triggered when certain criteria are met (Branke et al. [3]), however fixed schedules are much more popular and easier to analyze. Fernandez et al. studied the... |

35 | The island model genetic algorithm: On separability, population size and convergence
- Whitley, Rana, et al.
- 1999
(Show Context)
Citation Context ...migrants). There has been much research done on IMs. Theoretical studies of takeover times were published for example by Rudolph [9]. The number and size of populations were studied by Whitley et al. =-=[14]-=-. Cantú-Paz studied various parameters of island models (among others the population size, the connectivity, the migration rate, the takeover time and the selection intensity), some of them on more th... |

24 |
L.V.: An empirical study of multipopulation genetic programming
- Tomassini
- 2003
(Show Context)
Citation Context ...), however fixed schedules are much more popular and easier to analyze. Fernandez et al. studied the population number and size, the topology and the migration size and frequency in the context of GP =-=[6]-=-. Alba and Troya studied the asynchronism of island models [1]. The diversity of populations was studied by Ursem [13] and Gustafson [7]. Different representations, operators, strategies, etc. within ... |

24 | On takeover times in spatially structured populations: Array and ring
- Rudolph
- 2000
(Show Context)
Citation Context ...uency, and the emigration/immigration policy (how we choose and insert migrants). There has been much research done on IMs. Theoretical studies of takeover times were published for example by Rudolph =-=[9]-=-. The number and size of populations were studied by Whitley et al. [14]. Cantú-Paz studied various parameters of island models (among others the population size, the connectivity, the migration rate,... |

23 | An Analysis of Diversity in Genetic Programming
- Gustafson
- 2004
(Show Context)
Citation Context ...ology and the migration size and frequency in the context of GP [6]. Alba and Troya studied the asynchronism of island models [1]. The diversity of populations was studied by Ursem [13] and Gustafson =-=[7]-=-. Different representations, operators, strategies, etc. within one system are possible in heterogeneous IMs, and they help overcome many artificial constraints imposed by choices regarding the solvin... |

23 | Neutrality: A necessity for self-adaptation
- Toussaint, Igel
- 2002
(Show Context)
Citation Context ...posed by choices regarding the solving method. Changes in the exploration distribution around a given solution due to different representations or genetic operators are described by neutrality theory =-=[12]-=-. An interesting application of a heterogeneous setup is Injection Island GA [5] or a related Hierarchical Fair Competition model [8]. Switching representations dynamically (by ”shifting” Gray encodin... |

18 | Dynamic representations and escaping local optima: Improving genetic algorithms and local search
- Barbulescu, Watson, et al.
- 2000
(Show Context)
Citation Context ...ication of a heterogeneous setup is Injection Island GA [5] or a related Hierarchical Fair Competition model [8]. Switching representations dynamically (by ”shifting” Gray encodings) was presented in =-=[2]-=-. 3. RESEARCH DIRECTIONS The methodology used in the thesis will be primarily experimental. I will observe the changes in the behavior of an IM while changing its properties. 3.1 Test suite and measur... |

14 | The influence of migration sizes and intervals on island models
- Skolicki, Jong
- 2005
(Show Context)
Citation Context ...to study the negative influence of cooperation between islands (IM1), the positive influence of cooperation (IM2, IM3) and the influence of cooperation on overfitting to locally ideal solutions (IM4) =-=[11]-=-. Figure 4: IM3, average global diversity in runs with different migration intervals, migration size = 10 Another function was designed for studying heterogeneous IMs (F ). The function is difficult f... |

12 |
Hierarchical Fair Competition Model for Parallel Evolutionary Algorithms
- Hu, Goodman
(Show Context)
Citation Context ...entations or genetic operators are described by neutrality theory [12]. An interesting application of a heterogeneous setup is Injection Island GA [5] or a related Hierarchical Fair Competition model =-=[8]-=-. Switching representations dynamically (by ”shifting” Gray encodings) was presented in [2]. 3. RESEARCH DIRECTIONS The methodology used in the thesis will be primarily experimental. I will observe th... |

12 | Improving evolutionary algorithms with multi-representation island models
- Skolicki, Jong
- 2004
(Show Context)
Citation Context ... 10 Another function was designed for studying heterogeneous IMs (F ). The function is difficult for only binary or Gray encodings, but is much easier when switching between representation is allowed =-=[10]-=-. Additionally, standard multimodal functions, known as Rosenbrock, Rastrigin, Schwefel and Griewangk are also used. 4.2 Migration sizes and intervals Two very basic and important parameters of island... |

10 | An analysis of synchronous and asynchronous parallel distributed genetic algorithms with structured and panmictic islands
- Alba, Troya
- 1999
(Show Context)
Citation Context ... analyze. Fernandez et al. studied the population number and size, the topology and the migration size and frequency in the context of GP [6]. Alba and Troya studied the asynchronism of island models =-=[1]-=-. The diversity of populations was studied by Ursem [13] and Gustafson [7]. Different representations, operators, strategies, etc. within one system are possible in heterogeneous IMs, and they help ov... |

8 | The optimization of flywheels using an injection island genetic algorithm
- Eby, Averill, et al.
- 1999
(Show Context)
Citation Context ...ution around a given solution due to different representations or genetic operators are described by neutrality theory [12]. An interesting application of a heterogeneous setup is Injection Island GA =-=[5]-=- or a related Hierarchical Fair Competition model [8]. Switching representations dynamically (by ”shifting” Gray encodings) was presented in [2]. 3. RESEARCH DIRECTIONS The methodology used in the the... |

5 |
Distribution of evolutionary algorithms in heterogeneous networks
- Branke, Kamper, et al.
- 2004
(Show Context)
Citation Context ...election intensity), some of them on more theoreticalmodels and others in more experimental setups [4]. There exist models where migrations are triggered when certain criteria are met (Branke et al. =-=[3]-=-), however fixed schedules are much more popular and easier to analyze. Fernandez et al. studied the population number and size, the topology and the migration size and frequency in the context of GP ... |

5 |
Models for evolutionary Algorithms and Their
- Ursem
- 2003
(Show Context)
Citation Context ...r and size, the topology and the migration size and frequency in the context of GP [6]. Alba and Troya studied the asynchronism of island models [1]. The diversity of populations was studied by Ursem =-=[13]-=- and Gustafson [7]. Different representations, operators, strategies, etc. within one system are possible in heterogeneous IMs, and they help overcome many artificial constraints imposed by choices re... |