Loading...
Please wait, while we are loading the content...
Agent-Based CoEvolutionary Techniques for Solving Multi-Objective Optimization Problems
| Content Provider | Semantic Scholar |
|---|---|
| Author | Siwik, Leszek |
| Copyright Year | 2008 |
| Abstract | Evolutionary algorithms (EAs) are optimization and search techniques inspired by the Darwinian model of biological evolutionary processes (Bäck et al., 1997). EAs are robust and efficient techniques, which find approximate solutions to many problems which are difficult or even impossible to solve with the use of “classical” techniques. There are many different types of evolutionary algorithms developed during over 40 years of research. One of the branches of EAs are co-evolutionary algorithms (CEAs) (Paredis, 1998). The main difference between EAs and CEAs is the way in which the fitness of an individual is evaluated in each approach. In the case of evolutionary algorithms each individual has the solution of the given problem encoded within its genotype and its fitness depends only on how “good” is that solution. In the case of co-evolutionary algorithms of course there is also obviously solution to the given problem encoded within the individual’s genotype but the fitness is estimated on the basis of interactions of the given individual with other individuals present in the population. Thus co-evolutionary algorithms are applicable in the case of problems for which it is difficult or even impossible to formulate explicit fitness function—in such cases we can just encode the solutions within the individuals’ genotypes and individuals compete—or co-operate—with each other, and such process of interactions leads to the fitness estimation. Co-evolutionary interactions between individuals have also other positive effects. One of them is maintaining the population diversity, another one are “arms races”—continuous “progress” toward better and better solutions to the given problem via competition between species. Co-evolutionary algorithms are classified into two general categories: competitive and cooperative (Paredis, 1998). The main difference between these two types of co-evolutionary algorithms is the way in which the individuals interact during the fitness estimation. In the case of competitive co-evolutionary algorithms the value of fitness is estimated as a result of the series of tournaments, in which the individual for which the fitness is estimated and some other individuals from the population are engaged. The way of choosing the competitors for tournaments may vary in different versions of algorithms—for example it may be the competition with the best individual from the other species or competition with several randomly chosen individuals, etc. On the other hand, co-operative co-evolutionary algorithms (CCEAs) are CEAs in which there exist several sub-populations (species) (Potter & De Jong, 2000). Each of them solves O pe n A cc es s D at ab as e w w w .ite ch on lin e. co m |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://www.free-ebooks.net/academic-science/Agent-Based-Co-Evolutionary-Techniques-for-Solving-Multi-Objective-Optimization-Problems/pdf?dl=&preview= |
| Alternate Webpage(s) | http://home.agh.edu.pl/~drezew/papers/drezewski2008agent-based-coevolutionary.pdf |
| Alternate Webpage(s) | https://api.intechopen.com/chapter/pdf-download/5238 |
| Alternate Webpage(s) | https://www.researchgate.net/profile/Rafal_Drezewski/publication/221787332_Agent-Based_Co-Evolutionary_Techniques_for_Solving_Multi-Objective_Optimization_Problems/links/0912f512aa78c8073a000000.pdf |
| Alternate Webpage(s) | http://cdn.intechweb.org/pdfs/5238.pdf |
| Alternate Webpage(s) | http://www.intechweb.org/downloadpdf.php?id=5238 |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
