摘要
针对目前连续PH分布数据拟合EM(Expectation-Maximization)算法存在的初值敏感问题,提出运用确定性退火EM算法进行连续PH分布数据拟合,给出了详细的理论推导,并通过两个拟合实例与标准EM算法进行了对比.对比结果表明所提出的方法可以有效地避免初值选择的不同对EM算法结果的影响,减小陷入局部最优的可能性,能得到比标准EM算法更好的结果.
To overcome the initial parameters sensitive problem in the conventional expectation-maximization (EM) algorithm for phase-type distributions(PH) data fit, a PH distributions data fit method based on the deterministic annealing EM algorithm is proposed. The detailed theoretical inference process of the method is given. The method is compared with standard EM algorithm through two benchmarks. Contrast results show that the proposed method can obtain better estimates free of the initial parameter values and reduce the possibility of getting into local maximum.
出处
《控制与决策》
EI
CSCD
北大核心
2008年第2期133-139,共7页
Control and Decision
基金
国家自然科学基金项目(70501031)