KINDER: Kernel-based Independence for Fair Representation Learning via Prototype-space Erasure
Abstract
Concept erasure is a prominent approach to achieving fairness in machine learning through the removal of sensitive attributes from learned representations. Prior concept erasure methods typically define debiasing indirectly through the failure of a chosen adversary, probe, or fairness penalty, which makes the target of erasure dependent on a particular decoder family or optimization setup. We introduce KINDER, which directly removes sensitive-attribute information from intermediate representations with respect to a chosen family of prototype functionals, and is compatible with any deep model containing an intermediate feature space, independent of the architecture, task type, or training paradigm. KINDER operates in a random Fourier feature space, where it estimates a sensitive subspace from protected group prototypes and projects representations onto its orthogonal complement. Thus, it forms an explicit representation cleaning mechanism distinct from prior fairness control studies. We conduct extensive experiments across diverse settings and show the effectiveness of KINDER on (i) supervised and self-supervised settings, (ii) classification, regression, and image segmentation tasks, and (iii) diverse data modalities, including visual, textual, and tabular data. Code and trained models are available at: https://github.com/abtinmU/KINDER
Related
- Sequential Conditional Independence Testing with Machine Learning Models
- Self-normalized Bernstein Inequalities for Regenerative Processes and Applications to Learning Without Independence
- The Effectiveness of Discovery Learning on Mathematical Reflective Thinking Skills Based on the Level of Learning Independence of High School Students
- HyCoSeq: Contextual Hyperbolic Representation Learning for Genomic Sequences
- Representation Learning with Quantum Signal Processing
BibTeX
@misc{mogharabin2026kinder,
title = {{KINDER: Kernel-based Independence for Fair Representation Learning via Prototype-space Erasure}},
author = {Abtin Mogharabin and Alp Toykan Kaplan and Jiaee Cheong and Sinan Kalkan},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.00264},
eprint = {2611.00264},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.00264},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - KINDER: Kernel-based Independence for Fair Representation Learning via Prototype-space Erasure
AU - Abtin Mogharabin
AU - Alp Toykan Kaplan
AU - Jiaee Cheong
AU - Sinan Kalkan
PY - 2026
DA - 2026/11/01
PB - arXiv
DO - 10.48550/arxiv.2611.00264
UR - https://arxiv.org/abs/2611.00264
AB - Concept erasure is a prominent approach to achieving fairness in machine learning through the removal of sensitive attributes from learned representations. Prior concept erasure methods typically define debiasing indirectly through the failure of a chosen adversary, probe, or fairness penalty, which makes the target of erasure dependent on a particular decoder family or optimization setup. We introduce KINDER, which directly removes sensitive-attribute information from intermediate representations with respect to a chosen family of prototype functionals, and is compatible with any deep model containing an intermediate feature space, independent of the architecture, task type, or training paradigm. KINDER operates in a random Fourier feature space, where it estimates a sensitive subspace from protected group prototypes and projects representations onto its orthogonal complement. Thus, it forms an explicit representation cleaning mechanism distinct from prior fairness control studies. We conduct extensive experiments across diverse settings and show the effectiveness of KINDER on (i) supervised and self-supervised settings, (ii) classification, regression, and image segmentation tasks, and (iii) diverse data modalities, including visual, textual, and tabular data. Code and trained models are available at: https://github.com/abtinmU/KINDER
ER -