Evolutionary algorithms provide robust, relatively easy-to-use methods for a wide range of parameter estimation problems. Recent developments have led to more reliable and/or faster algorithms, and these methods may be considered along with the early generation evolutionary algorithms currently used in ecological modeling.
Evolutionary algorithms provide one of the few available methods to automatically generate models - rather than parameters - from data, their main competition being artificial neural networks (see Artificial Neural Networks) and relational learning systems. Their inherent white-box nature and robustness provide important advantages in automatic model generation, though much still remains to be done in these areas. They may be particularly appropriate where the underlying mechanisms of the process being modeled are evolution-like. In these cases, the application of coevolutionary techniques in ecological modeling merges seamlessly with the wider field of evolutionary simulation.
See also: Artificial Neural Networks; Ecological Models, Optimization; Spatial Models and Geographic Information Systems.
Was this article helpful?