Palestrante
Descrição
Genetic Algorithms (GAs) are non-deterministic search procedures based on the mechanisms of natural selection and genetics. They have been widely used for solving optimization problems in many contexts, such as physics, computer science and biology. An alternative probabilistic search procedures is based on social interactions like cooperation between the agents realizing the search process. In this context, we have the imitative learning algorithm that consists of a group of agents (represented by a set of strings) modifying its features (a bit of the string) to copy a random feature from the individual with highest fitness in the group. In a previous research (1), the performance of imitative learning algorithm was quantitatively, analyzed by measuring its average computational cost, which is defined as the product of the number of agents in the group and the number of trials they did until one of them finds the global maximum. Our goal in this research is to compare imitative learning performance with the performance of the sexual and asexual GAs. In order to do this, we will measure for both algorithms the average computational cost that is required for finding the global maximum in a rugged fitness landscape with varying complexity degrees generated using the NK model.
Referências
1 FONTANARI, J. F. Exploring NK fitness landscapes using imitative learning. European Physical Journal B, v. 88, p. 251-1-251-7, Oct. 2015. doi: 10.1140/epjb/e2015-60608-1.
| Subárea | Sistemas Complexos |
|---|---|
| Apresentação do trabalho acadêmico para o público geral | Sim |
