FALCON: False-Negative Aware Learning of Contrastive Negatives in Vision-Language Alignment

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kim, Myunsoo, Shim, Seongwoong, Lee, Byung-Jun
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912969859792896
author Kim, Myunsoo
Shim, Seongwoong
Lee, Byung-Jun
author_facet Kim, Myunsoo
Shim, Seongwoong
Lee, Byung-Jun
contents False negatives pose a critical challenge in vision-language pretraining (VLP) due to the many-to-many correspondence between images and texts in large-scale datasets. These false negatives introduce conflicting supervision signals that degrade the learned embedding space and diminish the effectiveness of hard negative sampling. In this paper, we propose FALCON (False-negative Aware Learning of COntrastive Negatives), a learning-based mini-batch construction strategy that adaptively balances the trade-off between hard and false negatives during VLP. Rather than relying on fixed heuristics, FALCON employs a negative mining scheduler that dynamically selects negative samples of appropriate hardness for each anchor instance during mini-batch construction, guided by a proxy for cross-modal alignment improvement. Experimental results demonstrate that FALCON significantly improves performance across three vision-language learning frameworks (ALBEF, BLIP-2, SigLIP-2) and a broad range of downstream tasks and evaluation settings, underscoring its effectiveness and robustness in mitigating the impact of false negatives.
format Preprint
id arxiv_https___arxiv_org_abs_2505_11192
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle FALCON: False-Negative Aware Learning of Contrastive Negatives in Vision-Language Alignment
Kim, Myunsoo
Shim, Seongwoong
Lee, Byung-Jun
Computer Vision and Pattern Recognition
Artificial Intelligence
False negatives pose a critical challenge in vision-language pretraining (VLP) due to the many-to-many correspondence between images and texts in large-scale datasets. These false negatives introduce conflicting supervision signals that degrade the learned embedding space and diminish the effectiveness of hard negative sampling. In this paper, we propose FALCON (False-negative Aware Learning of COntrastive Negatives), a learning-based mini-batch construction strategy that adaptively balances the trade-off between hard and false negatives during VLP. Rather than relying on fixed heuristics, FALCON employs a negative mining scheduler that dynamically selects negative samples of appropriate hardness for each anchor instance during mini-batch construction, guided by a proxy for cross-modal alignment improvement. Experimental results demonstrate that FALCON significantly improves performance across three vision-language learning frameworks (ALBEF, BLIP-2, SigLIP-2) and a broad range of downstream tasks and evaluation settings, underscoring its effectiveness and robustness in mitigating the impact of false negatives.
title FALCON: False-Negative Aware Learning of Contrastive Negatives in Vision-Language Alignment
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2505.11192