arXiv cs.AI· Kuniko Paxton, Koorosh Aslansefat, Dhavalkumar Thakker, Yiannis Papadopoulos·· 4 小时前AI 评分24
将肤色视为连续属性:皮肤病变分类中的肤色偏见评估与缓解
Evaluating and Mitigating Skin Color Bias in Skin Lesion Classificatio
AI 导读
一项研究提出将肤色表示为连续属性,并依据其与分类性能的非线性关系为每个样本分配权重,以缓解皮肤病变分类器中的肤色偏见。在两个皮肤病数据集上的评估显示,公平性指标对肤色分类方式敏感,该方法相比传统的分类重加权能带来更稳定的公平性提升,同时基本保持分类性能。
正文 · 原文
Abstract:Skin lesion classifiers may overlook performance disparity because of subtle skin tone variations by treating skin tone as a categorical attribute. This study proposes a method that represents skin tone as a continuous attribute and assigns individual weights to each sample based on its nonlinear relationship with classification performance. Evaluations on two dermatological datasets showed that fairness metrics were sensitive to skin-color categorization. The proposed method also demonstrated more stable fairness improvements than conventional categorical reweighting while generally preserving classification performance.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2506.11627 [cs.CV] |
| (or arXiv:2506.11627v2 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2506.11627 arXiv-issued DOI via DataCite |
Submission history
From: Kuniko Paxton [view email]
[v1]
Fri, 13 Jun 2025 09:54:01 UTC (3,555 KB)
[v2]
Tue, 17 Nov 2026 10:24:41 UTC (668 KB)
来源:arXiv cs.AI · arxiv.org