Class Unbiasing for Generalization in Medical Diagnosis

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zuo, Lishi, Mak, Man-Wai, Yi, Lu, Tu, Youzhi
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915472134373376
author Zuo, Lishi
Mak, Man-Wai
Yi, Lu
Tu, Youzhi
author_facet Zuo, Lishi
Mak, Man-Wai
Yi, Lu
Tu, Youzhi
contents Medical diagnosis might fail due to bias. In this work, we identified class-feature bias, which refers to models' potential reliance on features that are strongly correlated with only a subset of classes, leading to biased performance and poor generalization on other classes. We aim to train a class-unbiased model (Cls-unbias) that mitigates both class imbalance and class-feature bias simultaneously. Specifically, we propose a class-wise inequality loss which promotes equal contributions of classification loss from positive-class and negative-class samples. We propose to optimize a class-wise group distributionally robust optimization objective-a class-weighted training objective that upweights underperforming classes-to enhance the effectiveness of the inequality loss under class imbalance. Through synthetic and real-world datasets, we empirically demonstrate that class-feature bias can negatively impact model performance. Our proposed method effectively mitigates both class-feature bias and class imbalance, thereby improving the model's generalization ability.
format Preprint
id arxiv_https___arxiv_org_abs_2508_06943
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Class Unbiasing for Generalization in Medical Diagnosis
Zuo, Lishi
Mak, Man-Wai
Yi, Lu
Tu, Youzhi
Machine Learning
Artificial Intelligence
Medical diagnosis might fail due to bias. In this work, we identified class-feature bias, which refers to models' potential reliance on features that are strongly correlated with only a subset of classes, leading to biased performance and poor generalization on other classes. We aim to train a class-unbiased model (Cls-unbias) that mitigates both class imbalance and class-feature bias simultaneously. Specifically, we propose a class-wise inequality loss which promotes equal contributions of classification loss from positive-class and negative-class samples. We propose to optimize a class-wise group distributionally robust optimization objective-a class-weighted training objective that upweights underperforming classes-to enhance the effectiveness of the inequality loss under class imbalance. Through synthetic and real-world datasets, we empirically demonstrate that class-feature bias can negatively impact model performance. Our proposed method effectively mitigates both class-feature bias and class imbalance, thereby improving the model's generalization ability.
title Class Unbiasing for Generalization in Medical Diagnosis
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2508.06943