摘要
机器学习的普遍应用引起了人们对其信任度的关注,利用蜕变测试技术研究机器学习程序质量逐步成为软件测试领域的研究热点之一。为了对蜕变测试技术在机器学习领域的应用相关研究工作进行总体评价,并对蜕变测试技术进行了一些展望,首先,通过筛选得到多篇文献并对研究主题进行了分类,梳理了近年来蜕变测试在机器学习领域的应用和发展概况;其次,从应用场景、蜕变关系识别、蜕变测试度量和蜕变测试框架等方面分别对蜕变测试进行分析和讨论;最后,展望了蜕变测试在机器学习领域应用的未来研究热点。
The trust on machine learning has attracted people's attention along with its universal application,and the research on using metamorphic testing technology to ensure the quality of machine learning programs has gradually become one of the research hotspots in the field of software testing.The application of metamorphic testing technology in machine learning is summarised and some prospects of metamorphic testing technology are proposed.Firstly,the application and development of metamorphic testing in the field of machine learning is summarised through screening of multiple documents and classification of research topics.Then,the metamorphic testing is analyzed and discussed from the aspects of application scenario,metamorphic relation identification,metamorphic testing measurement,and metamorphic testing framework.Finally,the future research hotspots of the application of metamorphic testing in machine learning are prospected.
作者
潘娅
敖海洋
PAN Ya;AO Haiyang(School of Computer Science and Technology,Southwest University of Science and Technology,Mianyang 621010,China;Institute of Computer Application,Southwest University of Science and Technology,Mianyang 621010,China)
出处
《西安邮电大学学报》
2020年第4期81-89,共9页
Journal of Xi’an University of Posts and Telecommunications
基金
四川省重点研发计划项目(2020YFS0318,2019YFS0155)
四川省大学生创新创业训练计划项目(S202010619018)。
关键词
机器学习
机器学习测试
测试预言
蜕变测试
蜕变关系
machine learning
machine learning test
test oracle
metamorphic testing
metamorphic relationship