Evolutionary strategies and genetic algorithms for dynamic parameter optimization of evolving fuzzy neural networks

Evolving fuzzy neural networks are usually used to model evolving processes, which are developing and changing over time. This kind of network has some fixed parameters that usually depend on presented data. When data change over time, the best set of parameters also changes. This paper presents two approaches using evolutionary computation for the on-line optimization of these parameters. One of them utilizes genetic algorithms and the other one utilizes evolutionary strategies. The networks were used to Mackey-Glass chaotic time series prediction with changing dynamics. A comparative study is made with these approaches and some variations of them.