Host: Japan Society for Fuzzy Theory and Intelligent Informatics (SOFT)
In this paper, we have implemented some incremental classifier based on ARTMAP and we have compared the performance of these methods. There are two modified methods of ARTMAP that can take into account the degree of importance of attributes, ARTMAP-AW and ARTMAP-MD. ARTMAP-AW defines categories as hyper ellipses and it can ignore attributes irrelevant to classification on the basis of the importance attributes. In ARTMAP-MD, Mahalanobis distance is used to define the distance between a category and a case, so that it take into account not only the importance of attributes but also correlations between attributes. Although experimental results show that ARTMAP-AW acquires better classification rules than ARTMAP-MD, we have pointed out the modification to improve the performance of ARTMAP-MD.