Beyond Memorization: A Multi-Modal Ordinal Regression Benchmark to Expose Popularity Bias in Vision-Language Models

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Szu-Tu, Li-Zhong, Wu, Ting-Lin, Chang, Chia-Jui, Syu, He, Liu, Yu-Lun
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909975535681536
author Szu-Tu, Li-Zhong
Wu, Ting-Lin
Chang, Chia-Jui
Syu, He
Liu, Yu-Lun
author_facet Szu-Tu, Li-Zhong
Wu, Ting-Lin
Chang, Chia-Jui
Syu, He
Liu, Yu-Lun
contents We expose a significant popularity bias in state-of-the-art vision-language models (VLMs), which achieve up to 34% higher accuracy on famous buildings compared to ordinary ones, indicating a reliance on memorization over generalizable understanding. To systematically investigate this, we introduce the largest open benchmark for this task: the YearGuessr dataset, a collection of 55,546 building images with multi-modal attributes from 157 countries, annotated with continuous ordinal labels of their construction year (1001-2024), GPS data, and page-view counts as a proxy for popularity. Using this dataset, we frame the construction year prediction task as ordinal regression and introduce popularity-aware interval accuracy metrics to quantify this bias. Our resulting benchmark of 30+ models, including our YearCLIP model, confirms that VLMs excel on popular, memorized items but struggle significantly with unrecognized subjects, exposing a critical flaw in their reasoning capabilities. Project page: https://sytwu.github.io/BeyondMemo/
format Preprint
id arxiv_https___arxiv_org_abs_2512_21337
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Beyond Memorization: A Multi-Modal Ordinal Regression Benchmark to Expose Popularity Bias in Vision-Language Models
Szu-Tu, Li-Zhong
Wu, Ting-Lin
Chang, Chia-Jui
Syu, He
Liu, Yu-Lun
Computer Vision and Pattern Recognition
We expose a significant popularity bias in state-of-the-art vision-language models (VLMs), which achieve up to 34% higher accuracy on famous buildings compared to ordinary ones, indicating a reliance on memorization over generalizable understanding. To systematically investigate this, we introduce the largest open benchmark for this task: the YearGuessr dataset, a collection of 55,546 building images with multi-modal attributes from 157 countries, annotated with continuous ordinal labels of their construction year (1001-2024), GPS data, and page-view counts as a proxy for popularity. Using this dataset, we frame the construction year prediction task as ordinal regression and introduce popularity-aware interval accuracy metrics to quantify this bias. Our resulting benchmark of 30+ models, including our YearCLIP model, confirms that VLMs excel on popular, memorized items but struggle significantly with unrecognized subjects, exposing a critical flaw in their reasoning capabilities. Project page: https://sytwu.github.io/BeyondMemo/
title Beyond Memorization: A Multi-Modal Ordinal Regression Benchmark to Expose Popularity Bias in Vision-Language Models
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.21337