Tag
This paper proposes a hierarchical, interpretable-by-design pivot selection model based on proximity trees and ensemble learning. It is data-modality-agnostic and demonstrates competitive results across tabular, text, image, and time-series datasets.