Imbalanced ensemble learning leveraging a novel data-level diversity metric

Ying Pang, Lizhi Peng, Haibo Zhang, Zhenxiang Chen, Bo Yang

Abstract: Highlights•A data-level diversity measure IED to evaluate diversity without model training.•A new imbalanced ensemble learning model based on data-level diversity IED.•A new weight-adaptive voting strategy for ensemble learning.•Strong positive correlation between IED and classifier-based metrics.•Significant reduction on training time and improvement on learning performance.