An event-driven test framework for evolutionary algorithms in dynamic environments


Etaner-Uyar A., Uyar H.

Congress on Evolutionary Computation (CEC 2004), Oregon, United States Of America, 19 - 23 June 2004, pp.2265-2272 identifier

  • Publication Type: Conference Paper / Full Text
  • City: Oregon
  • Country: United States Of America
  • Page Numbers: pp.2265-2272

Abstract

There is a growing interest in applying evolutionary algorithms to dynamic environments. This can be seen by the increase in the number of papers addressing various issues concerning different types of changing environments. Most of these studies use their own test functions and implementations which makes it harder to repeat the reported experiments. This paper focuses on the types of benchmark functions found in literature that are used for testing evolutionary algorithms in dynamic environments and proposes an easy-to-use test framework. The proposed test framework consists of an event driven environment control mechanism component as well as a scenario creator component which allows the user to experiment with different instances of a problem, incorporating various properties and types of change. The components of the framework interact with each other through a well defined function call interface. By way of these features, the test framework realizes its design goals which are ease of use, compatibility, controllability, flexibility and generalizability. The test framework is still in its early stages of development. Extensions and modifications will be made based on the requirements and feedback received from its users.