Preliminary results on the application of boosting to learn weighted fuzzy rules under single-winner inference