Partial Information Decomposition for Data Interpretability and Feature Selection
Published in arXiv preprint, 2024
Abstract
The paper introduces PIDF, a novel approach that simultaneously addresses data interpretability and feature selection. Rather than assigning single importance scores, the method evaluates each feature using three metrics: the mutual information shared with the target variable, the feature’s contribution to synergistic information, and the amount of this information that is redundant. The authors validate their framework using synthetic and real-world datasets, with applications in genetics and neuroscience research.
Recommended citation: Westphal, C., Hailes, S., & Musolesi, M. (2024). Partial Information Decomposition for Data Interpretability and Feature Selection. arXiv preprint arXiv:2405.19212.
Download Paper