研究预训练图像模型在年龄、种族、性别交叉下的偏见，发现年龄区分最明显。
Assessing Intersectional Bias in Representations of Pre-Trained Image Recognition Models
- 用线性分类器探测和拓扑图可视化模型表征中的偏见
- 年龄差异在表征中区分度最强，种族与性别关联较弱
- 揭示了模型对特定群体的潜在歧视，适合关注公平性的研究者
深度学习模型取得显著成功，其训练常依赖预训练模型，但可能延续编码中的偏见。本文研究常用ImageNet分类器在人脸图像上的表征偏见，同时考虑年龄、种族和性别等敏感变量的交叉影响。通过线性分类器探针和激活拓扑图可视化，发现这些模型表征中对年龄的区分能力尤为突出；对特定族裔的关联较弱，且在中年群体中可辨识性别差异。
原文摘要 · Abstract (English)
Deep Learning models have achieved remarkable success. Training them is often accelerated by building on top of pre-trained models which poses the risk of perpetuating encoded biases. Here, we investigate biases in the representations of commonly used ImageNet classifiers for facial images while considering intersections of sensitive variables age, race and gender. To assess the biases, we use linear classifier probes and visualize activations as topographic maps. We find that representations in ImageNet classifiers particularly allow differentiation between ages. Less strongly pronounced, the models appear to associate certain ethnicities and distinguish genders in middle-aged groups.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。