ADAPTATION SCHEMES FOR A NEURO-FUZZY PREDICTOR
Lavinia Eugenia Ferariu, Horia Nicolai Teodorescu, Lucian Iulian Fira · 2005
The paper presents, in a comparative study, six hybrid configurations that can assist the adaptation of the neuro-fuzzy system parameters. The suggested configurations involve various combinations of the gradient descent algorithm, the k-means clustering procedure and the genetic algorithms. The neurofuzzy system architecture is adopted from literature. The applicability of the methodology is investigated with respect to the time series prediction.