李轶,蔡天训,樊建峰,吴文渊,冯勇.基于SVM的多项式循环程序秩函数生成.软件学报,2019,30(7):1903-1915 |
基于SVM的多项式循环程序秩函数生成 |
SVM-based Method for Detecting Ranking Functions in Polynomial Loop Programs |
投稿时间:2018-07-10 修订日期:2018-09-28 |
DOI:10.13328/j.cnki.jos.005748 |
中文关键词: 程序终止性 SVM 机器学习 秩函数 |
英文关键词:program termination SVM machine learning ranking functions |
基金项目:国家自然科学基金(61572024,61103110,11471307) |
|
摘要点击次数: 2085 |
全文下载次数: 999 |
中文摘要: |
程序终止性问题是自动程序验证领域中的一个研究热点.秩函数探测是进行终止性分析的主要方法.针对单重无条件分支的多项式循环程序,将其秩函数计算问题归结为二分类问题,从而可利用支持向量机(SVM)算法来计算程序的秩函数.与基于量词消去技术的秩函数计算方法不同,该方法能在可接受的时间范围内探测到更为复杂的秩函数. |
英文摘要: |
Synthesizing ranking functions of polynomial loop programs is the dominant method for checking their termination. In this study, the synthesis of ranking functions of a class of polynomial loop program is reduced to the binary problem. The support vector machine (SVM) technique then is applied to solve such the binary problem. This naturally relates detection of ranking functions to SVM. Different from the CAD-based method for synthesizing ranking functions, the proposed method can get more expressive polynomial ranking functions in an acceptable time. |
HTML 下载PDF全文 查看/发表评论 下载PDF阅读器 |