Queue-based Resampling for Online Class Imbalance Learning

TL;DR

This paper introduces queue-based resampling, a novel method for online class imbalance learning that effectively handles both class imbalance and concept drift by selectively including past examples, improving learning speed and quality.

Contribution

The paper presents a new queue-based resampling algorithm specifically designed for online class imbalance learning with concept drift, addressing a gap in existing research.

Findings

Outperforms state-of-the-art methods in benchmark tests

Enhances learning speed and quality in imbalanced data streams

Effectively manages concept drift with selective past example inclusion

Abstract

Online class imbalance learning constitutes a new problem and an emerging research topic that focusses on the challenges of online learning under class imbalance and concept drift. Class imbalance deals with data streams that have very skewed distributions while concept drift deals with changes in the class imbalance status. Little work exists that addresses these challenges and in this paper we introduce queue-based resampling, a novel algorithm that successfully addresses the co-existence of class imbalance and concept drift. The central idea of the proposed resampling algorithm is to selectively include in the training set a subset of the examples that appeared in the past. Results on two popular benchmark datasets demonstrate the effectiveness of queue-based resampling over state-of-the-art methods in terms of learning speed and quality.