An Empirical Study of the Imbalance Issue in Software Vulnerability Detection

Yuejun Guo ID, Qiang Hu ID, Qiang Tang ID, Yves Le Traon ID · 2026

Read An Empirical Study of the Imbalance Issue in Software Vulnerability Detection by Yuejun Guo ID, Qiang Hu ID, Qiang Tang ID, Yves Le Traon ID (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-12038.txt.

Description

Deep learning promises automated vulnerability detection, yet performance swings across datasets for reasons that were unclear. Yuejun Guo and three co-authors conjecture a culprit borrowed from computer vision: extreme class imbalance, since vulnerable code is a tiny fraction of all code. They test that idea in a broad empirical study across nine open-source datasets and two state-of-the-art deep learning models, confirming the hunch.

Existing fixes behave unevenly across test sets and metrics, the study finds. Focal loss best improves hit quality, mean false error and class-balanced loss lift recall, and random over-sampling aids the F1 measure, but none excels everywhere. The authors also probe outside effects on these fixes and offer guidance for building new ones. For flaw hunting, skew is not a side issue but the central obstacle.

Catalog record

CREATORYuejun Guo ID, Qiang Hu ID, Qiang Tang ID, Yves Le Traon ID
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE42 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-12038

Browse the full library · About the Tower of Babel