Generation and improvement of fuzzy classifiers with incremental learning using fuzzy RuleNet

Incremental learning is useful in applications where not all of the learning examples are available at the training time or where the system should remain adaptive during its operation. In this paper the influence of the incremental learning strategy on the performance of the fury-neuro model Fuzzy RuleNet is analyzed. For this purpose, Fuzzy RuleNet is presented shortly. Further, the applicability of the incremental learning technique is examined at a pattern recognition problem, namely the locahzation of address labels on postal parcels.