摘要
以提高验证效率、缩短验证周期为目标,使用贝叶斯网络优化模拟矢量,有效地缩小了用于回归测试的模拟矢量规模.采用信息论中的互信息作为评测准则,在输入变量和分支语句之间建立贝叶斯网络,并使用该网络进行推理和产生新的模拟矢量.实验结果表明:使用不同推理算法生成的模拟矢量长度大大缩短,平均为原有模拟矢量的1/10左右,其中最高路径覆盖率达到甚至超过了原有样本.
Improving the efficiency of simulation-based for regression test are huge and unnecessary. They made validation is important. Most of simulation vectors the covering process inefficient. In this paper, we used Bayesian network to describe the relation between the inputs and the branch statements. The new simulation vectors were generated by reasoning on the network. We performed experiments on some functional modules. The results indicate that the average vector length generated by the Bayesian network using different reference algorithms is about 10% of the original one, but the best path coverage even exceeds the original one.
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2007年第5期616-621,共6页
Journal of Computer-Aided Design & Computer Graphics
基金
国家"八六三"高技术研究发展计划(2004AA1Z1010).
关键词
贝叶斯网络
机器学习
路径覆盖率
模拟矢量生成
验证
validation Bayesian network
machine learning
path coverage
generation of simulation vectors