CL-ECDD: A contrastive learning framework for enterprise-level code defect detection

Liyuan Xue

doi:10.1016/j.neucom.2026.133298

CL-ECDD: A contrastive learning framework for enterprise-level code defect detection

2026-03-09

0

PRE

AI

DOI：10.1016/j.neucom.2026.133298

Original

Original request for help

Save

Abstract

En 中文

• Proposed a CL-ECDD framework, a contrastive learning framework for enterprise-level code defect detection. • A hard negative sample pool improves contrastive learning effectiveness and stability. • Multi-task learning with functional clustering enhances semantic representation learning. • Self-attention captures local code fragment relations for finer-grained defect detection. • The model achieves 97.72% precision and 96.41% recall on industrial real-world data.

Keywords:

contrastive learning

code defect detection

multi-task learning

self-attention

enterprise-level software

CL-ECDD: A contrastive learning framework for enterprise-level code defect detection

Abstract

Journal

Organization

Cited Papers