Conformal Recursive Feature Elimination
- 1Institute of Data Science and Artificial Intelligence (DATAI), University of Navarra, Ismael Sánchez Bella Building, Campus Universitario, Pamplona, 31009, Navarra, Spain
- 2TECNUN School of Engineering, Universidad de Navarra, Donostia-San Sebastián, 20018, Basque Country, Spain
Abstract
In this work, we introduce a novel feature selection method that leverages the conformal prediction framework. This new method, named Conformal Recursive Feature Elimination (CRFE), recursively identifies and removes features that increase the non-conformity of a dataset measured under additively decomposable non-conformity functions, re-estimating feature relevance after each elimination step analogously to classical Recursive Feature Elimination (RFE). We also introduce a geometrically motivated stopping criterion for recursive feature selectors. CRFE outputs are compared with the classical RFE algorithm and the results of other state-of-the-art feature selectors across several benchmark and real-world datasets. Our experiments show that CRFE generally improves prediction-set efficiency and subset stability relative to RFE, while maintaining empirical coverage close to the nominal level. Finally, we show how the automatic stopping criterion effectively halted the recursive selection of features, selecting subsets of effective and non-redundant features. • A novel feature selector named CRFE is presented. • CRFE leverages conformal prediction to recursively remove features. • CRFE improves standard RFE in terms of performance and stability. • A data-driven stopping criterion is presented and validated.
Related
- Exact Dynamics and Finite-Sample Trajectory Recovery of Linear Recursive Feature Machines
- Sequential Conformal Calibration under Recursive Selection
- Large Growth Happens: Gaussian Elimination with Partial Pivoting on Random Matrices
- Elimination of René Peschmann's 968 Remaining Hard Fibers for the Perfect Cuboid Problem
- Accelerating Fourier--Motzkin elimination: redundancy removal and the choice of variable elimination order
BibTeX
@article{lopezdecastro2026conformal,
title = {{Conformal Recursive Feature Elimination}},
author = {Marcos López-De-Castro and Alberto García‐Galindo and Rubén Armañanzas},
journal = {Pattern Recognition},
year = {2026},
month = jul,
doi = {10.1016/j.patcog.2026.114450},
url = {https://doi.org/10.1016/j.patcog.2026.114450},
note = {CC BY-NC-ND},
}
RIS
TY - JOUR
TI - Conformal Recursive Feature Elimination
AU - Marcos López-De-Castro
AU - Alberto García‐Galindo
AU - Rubén Armañanzas
PY - 2026
DA - 2026/07/10
JO - Pattern Recognition
DO - 10.1016/j.patcog.2026.114450
UR - https://doi.org/10.1016/j.patcog.2026.114450
AB - In this work, we introduce a novel feature selection method that leverages the conformal prediction framework. This new method, named Conformal Recursive Feature Elimination (CRFE), recursively identifies and removes features that increase the non-conformity of a dataset measured under additively decomposable non-conformity functions, re-estimating feature relevance after each elimination step analogously to classical Recursive Feature Elimination (RFE). We also introduce a geometrically motivated stopping criterion for recursive feature selectors. CRFE outputs are compared with the classical RFE algorithm and the results of other state-of-the-art feature selectors across several benchmark and real-world datasets. Our experiments show that CRFE generally improves prediction-set efficiency and subset stability relative to RFE, while maintaining empirical coverage close to the nominal level. Finally, we show how the automatic stopping criterion effectively halted the recursive selection of features, selecting subsets of effective and non-redundant features. • A novel feature selector named CRFE is presented. • CRFE leverages conformal prediction to recursively remove features. • CRFE improves standard RFE in terms of performance and stability. • A data-driven stopping criterion is presented and validated.
SN - 0031-3203
SN - 1873-5142
ER -