Enhanced fuzzy relational classifier with representative training samples

Fuzzy relational classifier (FRC) has been proven effective in both revealing the data structure and interpreting classification result. In FRC, fuzzy matrix R describing the relationship between the clusters and class labels plays an important role in its effective and robust classification. However, original FRC employs all the training samples undifferentiatedly to construct R, and thus leading to three disadvantages: lack of robustness for classification, degeneration on the classification performance and high computational load. To overcome these disadvantages, in this paper, a simple Enhanced Fuzzy Relational Classifier (EFRC) is developed by employing the training samples differentiatedly to build a more robust and effective R. Experimental results show that the proposed EFRC performs effectively and efficiently on both artificial and real datasets.