Related Experiment Videos
An alternative explanation for the manner in which genetic algorithms operate
1University of Dortmund, Department of Computer Science, Germany. beyer@LS 11.informatik.uni-dortmund.de
Bio Systems
|January 1, 1997
Summary
Genetic algorithms (GAs) may not rely solely on combining solution parts. Instead, collective phenomena in populations undergoing recombination offer an alternative explanation for their success, applicable to all evolutionary algorithms.
Area of Science:
- Computational Intelligence
- Evolutionary Computation
- Bio-inspired Computing
Background:
- The 'building block hypothesis' is the prevailing explanation for genetic algorithm (GA) efficacy.
- This hypothesis posits that GAs improve solutions by combining 'building blocks' from existing ones.
Purpose of the Study:
- To present an alternative explanation for how genetic algorithms (GAs) and other evolutionary algorithms (EAs) process populations.
- To explore collective phenomena during recombination as a key driver of evolutionary progress.
Main Methods:
- Analysis of principles from evolution strategies (ESs).
- Generalization of these principles to all evolutionary algorithms (EAs), including genetic algorithms (GAs).
Main Results:
- An alternative model focusing on collective recombination dynamics is proposed.
- The findings suggest that emergent population-level behaviors are crucial for evolutionary success.
Conclusions:
- The study offers a new perspective on evolutionary algorithms, shifting focus from individual component combination to population-level dynamics.
- The principles identified are general and may offer insights analogous to the benefits of sexual reproduction in biological systems.