摘要
提出一种基于自动搜索的概率分布生成方法,设计对概率分布的表示形式与评估函数,同时结合模拟退火算法设计基于马尔可夫模型的自动搜索过程。实验结果表明,该方法能够有效地提高自动搜索的准确性,在一定时间内成功找到接近最优的概率分布,生成高效的测试数据,同时达到降低统计测试成本的目的。
Astrategy based on automated search for probability distribution construction is proposed,which comprises the design of representation format and evaluation function for the probability distribution.Combining with simulated annealing algorithm,an indicator is defined to formalize the automated search process based on the Markov model.Experimental results show that the method effectively improves the accuracy of the automated search,which can reduce the expense of statistical test by providing the statistical test with fairly efficient test data since it successfully finds the neat-optimal probability distribution within a certain time.
出处
《计算机工程》
CAS
CSCD
2013年第2期274-278,共5页
Computer Engineering
基金
国家自然科学基金资助项目(61073163)
上海市科委基金资助项目(09220503000)
上海市引进技术的吸收与创新计划基金资助项目(2010CH-014)
关键词
统计测试
概率分布
自动搜索
覆盖约束
模拟退火算法
马尔可夫模型
statistical test
probability distribution
automated search
coverage constraint
simulated annealing algorithm
Markov model
