Aggregation in Ensemble Neural Models with Type-3 and Intuitionistic Fuzzy Logic

Type-3 intuitionistic fuzzy aggregation in ensembles of neural networks is offered in this work. Aggregation is essential for combining the prediction of the networks that constitute the ensemble. The fuzzy aggregator is designed such that the combined prediction is better than the individual predictions. The type-3 intuitionistic fuzzy system estimates the weights assigned to the individual predictions in a weighted average approach. Prediction uncertainty based on data is modeled by type-3 and uncertainty of the experts is modeled by intuitionistic fuzzy sets. The proposal is tested with an illustrative example of prediction.