Assessing Intersectional Bias in Representations of Pre-Trained Image Recognition Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Krug, Valerie, Stober, Sebastian
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913880681218048
author Krug, Valerie
Stober, Sebastian
author_facet Krug, Valerie
Stober, Sebastian
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2506_03664
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Assessing Intersectional Bias in Representations of Pre-Trained Image Recognition Models
Krug, Valerie
Stober, Sebastian
Computer Vision and Pattern Recognition
Computers and Society
Human-Computer Interaction
Machine Learning
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.
title Assessing Intersectional Bias in Representations of Pre-Trained Image Recognition Models
topic Computer Vision and Pattern Recognition
Computers and Society
Human-Computer Interaction
Machine Learning
url https://arxiv.org/abs/2506.03664