A Fuzzy Variant for On-Demand Data Stream Classification

Abstract

In many real-world applications, data arrive sequentially in the form of streams. Processing such data poses challenges to machine learning. In data streams learning, classification problems aim to predict the true class of incoming instances in real time. While adhering to online learning strategies, in this paper we extend the On-Demand classification algorithm to include concepts of fuzzy sets theory as a way to make classification more flexible to stream changes. A set of experiments was conducted to evaluate the proposed method. Experiments show that our approach is promising in dealing with imbalanced data streams and presents benefits with relation to the non-fuzzy version.