深度学习在软件缺陷检测交叉实验中的应用研究  

Research on the Application of Deep Learning in Cross Experiment of Software Defect Detection

作　　者：李杭桔 LI Hangju(Sichuan University,Chengdu Sichuan 610207)

机构地区：[1]四川大学,四川成都610207

出　　处：《软件》2025年第3期153-155,共3页Software

摘　　要：随着5G通信技术的迅速发展,通信软件系统日益庞大且复杂,传统缺陷检测方法在效率与准确性方面难以满足高质量的开发需求。深度学习凭借强大的特征提取与模式识别能力,通过特定的代码表示方法与深度神经网络架构设计,能够有效识别通信软件中的功能性缺陷与性能缺陷。该方法在降低了人工检测成本的同时,显著提升了缺陷识别准确率,对推动通信软件工程技术创新具有重要意义。With the rapid development of 5G communication technology,communication software systems are becoming increasingly large and complex,and traditional defect detection methods are unable to meet the requirements of high-quality development in terms of efficiency and accuracy.Deep learning,with its powerful feature extraction and pattern recognition capabilities,can effectively identify functional and performance defects in communication software through specific code representation methods and deep neural network architecture design.This method significantly improves the accuracy of defect recognition while reducing the cost of manual inspection,which is of great significance for promoting innovation in communication software engineering technology.

关 键 词：深度学习 软件缺陷检测 交叉实验 迁移学习 领域适应 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TP311.13[自动化与计算机技术—控制科学与工程]