Abstract

Rough-Fuzzy Support Vector Data Description is a novel soft computing derivative of the classical Support Vector Data Description algorithm used in many real-world applications successfully. However, its current version treats all data points equally when constructing the classifier. If the data set contains outliers, they will substantially affect the decision boundary. To overcome this issue, we present a novel approach based on the induced ordered weighted average operator and linguistic quantifier functions to weigh data points depending on their closeness to the lower approximation of the target class. In this way, we determine the weights for the data points without using any external procedure. Our computational experiments emphasize the strength of the proposed approach underlining its potential for outlier detection.