Guardado en:
Detalles Bibliográficos
Autores principales: Jin, Gaojie, Wu, Sihao, Liu, Jiaxu, Huang, Tianjin, Mu, Ronghui
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2501.13273
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866909464248975360
author Jin, Gaojie
Wu, Sihao
Liu, Jiaxu
Huang, Tianjin
Mu, Ronghui
author_facet Jin, Gaojie
Wu, Sihao
Liu, Jiaxu
Huang, Tianjin
Mu, Ronghui
contents Recent research has highlighted a critical issue known as ``robust fairness", where robust accuracy varies significantly across different classes, undermining the reliability of deep neural networks (DNNs). A common approach to address this has been to dynamically reweight classes during training, giving more weight to those with lower empirical robust performance. However, we find there is a divergence of class-wise robust performance between training set and testing set, which limits the effectiveness of these explicit reweighting methods, indicating the need for a principled alternative. In this work, we derive a robust generalization bound for the worst-class robust error within the PAC-Bayesian framework, accounting for unknown data distributions. Our analysis shows that the worst-class robust error is influenced by two main factors: the spectral norm of the empirical robust confusion matrix and the information embedded in the model and training set. While the latter has been extensively studied, we propose a novel regularization technique targeting the spectral norm of the robust confusion matrix to improve worst-class robust accuracy and enhance robust fairness. We validate our approach through comprehensive experiments on various datasets and models, demonstrating its effectiveness in enhancing robust fairness.
format Preprint
id arxiv_https___arxiv_org_abs_2501_13273
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Enhancing Robust Fairness via Confusional Spectral Regularization
Jin, Gaojie
Wu, Sihao
Liu, Jiaxu
Huang, Tianjin
Mu, Ronghui
Machine Learning
Recent research has highlighted a critical issue known as ``robust fairness", where robust accuracy varies significantly across different classes, undermining the reliability of deep neural networks (DNNs). A common approach to address this has been to dynamically reweight classes during training, giving more weight to those with lower empirical robust performance. However, we find there is a divergence of class-wise robust performance between training set and testing set, which limits the effectiveness of these explicit reweighting methods, indicating the need for a principled alternative. In this work, we derive a robust generalization bound for the worst-class robust error within the PAC-Bayesian framework, accounting for unknown data distributions. Our analysis shows that the worst-class robust error is influenced by two main factors: the spectral norm of the empirical robust confusion matrix and the information embedded in the model and training set. While the latter has been extensively studied, we propose a novel regularization technique targeting the spectral norm of the robust confusion matrix to improve worst-class robust accuracy and enhance robust fairness. We validate our approach through comprehensive experiments on various datasets and models, demonstrating its effectiveness in enhancing robust fairness.
title Enhancing Robust Fairness via Confusional Spectral Regularization
topic Machine Learning
url https://arxiv.org/abs/2501.13273