Guardado en:
Detalles Bibliográficos
Autores principales: Matos, Ivan Luiz De Moura, Saoud, Abdel Djalil Sad, Iakovleva, Ekaterina, Pastore, Vito Paolo, Tartaglione, Enzo
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:https://arxiv.org/abs/2603.05582
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916006163644416
author Matos, Ivan Luiz De Moura
Saoud, Abdel Djalil Sad
Iakovleva, Ekaterina
Pastore, Vito Paolo
Tartaglione, Enzo
author_facet Matos, Ivan Luiz De Moura
Saoud, Abdel Djalil Sad
Iakovleva, Ekaterina
Pastore, Vito Paolo
Tartaglione, Enzo
contents The issue of algorithmic biases in deep learning has led to the development of various debiasing techniques, many of which perform complex training procedures or dataset manipulation. However, an intriguing question arises: is it possible to extract fair and bias-agnostic subnetworks from standard vanilla-trained models without relying on additional data, such as unbiased training set? In this work, we introduce Bias-Invariant Subnetwork Extraction (BISE), a learning strategy that identifies and isolates "bias-free" subnetworks that already exist within conventionally trained models, without retraining or finetuning the original parameters. Our approach demonstrates that such subnetworks can be extracted via pruning and can operate without modification, effectively relying less on biased features and maintaining robust performance. Our findings contribute towards efficient bias mitigation through structural adaptation of pre-trained neural networks via parameter removal, as opposed to costly strategies that are either data-centric or involve (re)training all model parameters. Extensive experiments on common benchmarks show the advantages of our approach in terms of the performance and computational efficiency of the resulting debiased model.
format Preprint
id arxiv_https___arxiv_org_abs_2603_05582
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Bias In, Bias Out? Finding Unbiased Subnetworks in Vanilla Models
Matos, Ivan Luiz De Moura
Saoud, Abdel Djalil Sad
Iakovleva, Ekaterina
Pastore, Vito Paolo
Tartaglione, Enzo
Machine Learning
Computer Vision and Pattern Recognition
The issue of algorithmic biases in deep learning has led to the development of various debiasing techniques, many of which perform complex training procedures or dataset manipulation. However, an intriguing question arises: is it possible to extract fair and bias-agnostic subnetworks from standard vanilla-trained models without relying on additional data, such as unbiased training set? In this work, we introduce Bias-Invariant Subnetwork Extraction (BISE), a learning strategy that identifies and isolates "bias-free" subnetworks that already exist within conventionally trained models, without retraining or finetuning the original parameters. Our approach demonstrates that such subnetworks can be extracted via pruning and can operate without modification, effectively relying less on biased features and maintaining robust performance. Our findings contribute towards efficient bias mitigation through structural adaptation of pre-trained neural networks via parameter removal, as opposed to costly strategies that are either data-centric or involve (re)training all model parameters. Extensive experiments on common benchmarks show the advantages of our approach in terms of the performance and computational efficiency of the resulting debiased model.
title Bias In, Bias Out? Finding Unbiased Subnetworks in Vanilla Models
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.05582