A new parallel expectation-maximization （EM） algorithm is proposed for large databases. The purpose of the algorithm is to accelerate the operation of the EM algorithm. As a well-known algorithm for estimation in ge...A new parallel expectation-maximization （EM） algorithm is proposed for large databases. The purpose of the algorithm is to accelerate the operation of the EM algorithm. As a well-known algorithm for estimation in generic statistical problems, the EM algorithm has been widely used in many domains. But it often requires significant computational resources. So it is needed to develop more elaborate methods to adapt the databases to a large number of records or large dimensionality. The parallel EM algorithm is based on partial Esteps which has the standard convergence guarantee of EM. The algorithm utilizes fully the advantage of parallel computation. It was confirmed that the algorithm obtains about 2.6 speedups in contrast with the standard EM algorithm through its application to large databases. The running time will decrease near linearly when the number of processors increasing.展开更多
Flight delay prediction remains an important research topic due to dynamic nature in flight operation and numerous delay factors.Dynamic data-driven application system in the control area can provide a solution to thi...Flight delay prediction remains an important research topic due to dynamic nature in flight operation and numerous delay factors.Dynamic data-driven application system in the control area can provide a solution to this problem.However,in order to apply the approach,a state-space flight delay model needs to be established to represent the relationship among system states,as well as the relationship between system states and input/output variables.Based on the analysis of delay event sequence in a single flight,a state-space mixture model is established and input variables in the model are studied.Case study is also carried out on historical flight delay data.In addition,the genetic expectation-maximization（EM）algorithm is used to obtain the global optimal estimates of parameters in the mixture model,and results fit the historical data.At last,the model is validated in Kolmogorov-Smirnov tests.Results show that the model has reasonable goodness of fitting the data,and the search performance of traditional EM algorithm can be improved by using the genetic algorithm.展开更多
EM算法是近年来常用的求后验众数的估计的一种数据增广算法,但由于求出其E步中积分的显示表达式有时很困难,甚至不可能,限制了其应用的广泛性.而Monte Carlo EM算法很好地解决了这个问题,将EM算法中E步的积分用Monte Carlo模拟来有效实...EM算法是近年来常用的求后验众数的估计的一种数据增广算法,但由于求出其E步中积分的显示表达式有时很困难,甚至不可能,限制了其应用的广泛性.而Monte Carlo EM算法很好地解决了这个问题,将EM算法中E步的积分用Monte Carlo模拟来有效实现,使其适用性大大增强.但无论是EM算法,还是Monte Carlo EM算法,其收敛速度都是线性的,被缺损信息的倒数所控制,当缺损数据的比例很高时,收敛速度就非常缓慢.而Newton-Raphson算法在后验众数的附近具有二次收敛速率.本文提出Monte Carlo EM加速算法,将Monte Carlo EM算法与Newton-Raphson算法结合,既使得EM算法中的E步用Monte Carlo模拟得以实现,又证明了该算法在后验众数附近具有二次收敛速度.从而使其保留了Monte Carlo EM算法的优点,并改进了Monte Carlo EM算法的收敛速度.本文通过数值例子,将Monte Carlo EM加速算法的结果与EM算法、Monte Carlo EM算法的结果进行比较,进一步说明了Monte Carlo EM加速算法的优良性.展开更多
提出一种基于贪心EM算法的HMRF遥感影像变化检测算法。该算法采取PCA与差值法相结合的方式来构造差分影像。首先,采用隐马尔可夫随机场(Hidden Markov Random Field,HMRF)模型描述空间上下文信息,并构造系统能量函数;然后,利用贪心EM算...提出一种基于贪心EM算法的HMRF遥感影像变化检测算法。该算法采取PCA与差值法相结合的方式来构造差分影像。首先,采用隐马尔可夫随机场(Hidden Markov Random Field,HMRF)模型描述空间上下文信息,并构造系统能量函数;然后,利用贪心EM算法克服EM算法假定混合成分数为已知、迭代结果过分依赖初始值、可能收敛到局部最大点或收敛到参数空间边界的缺点,能够准确学习分布模型结构和参数,发现数据对模型的最佳匹配;最后,通过条件迭代模型(Iterated Conditional Modes,ICM)优化算法求解能量函数最优解,获取变化区域。实验结果表明,该算法能够更好地保持影像的结构性,有效去除孤立噪声。展开更多
基金the National Natural Science Foundation of China(79990584)
文摘A new parallel expectation-maximization （EM） algorithm is proposed for large databases. The purpose of the algorithm is to accelerate the operation of the EM algorithm. As a well-known algorithm for estimation in generic statistical problems, the EM algorithm has been widely used in many domains. But it often requires significant computational resources. So it is needed to develop more elaborate methods to adapt the databases to a large number of records or large dimensionality. The parallel EM algorithm is based on partial Esteps which has the standard convergence guarantee of EM. The algorithm utilizes fully the advantage of parallel computation. It was confirmed that the algorithm obtains about 2.6 speedups in contrast with the standard EM algorithm through its application to large databases. The running time will decrease near linearly when the number of processors increasing.
基金Supported by the High Technology Research and Development Programme of China(2006AA12A106)~~
文摘Flight delay prediction remains an important research topic due to dynamic nature in flight operation and numerous delay factors.Dynamic data-driven application system in the control area can provide a solution to this problem.However,in order to apply the approach,a state-space flight delay model needs to be established to represent the relationship among system states,as well as the relationship between system states and input/output variables.Based on the analysis of delay event sequence in a single flight,a state-space mixture model is established and input variables in the model are studied.Case study is also carried out on historical flight delay data.In addition,the genetic expectation-maximization（EM）algorithm is used to obtain the global optimal estimates of parameters in the mixture model,and results fit the historical data.At last,the model is validated in Kolmogorov-Smirnov tests.Results show that the model has reasonable goodness of fitting the data,and the search performance of traditional EM algorithm can be improved by using the genetic algorithm.
文摘EM算法是近年来常用的求后验众数的估计的一种数据增广算法,但由于求出其E步中积分的显示表达式有时很困难,甚至不可能,限制了其应用的广泛性.而Monte Carlo EM算法很好地解决了这个问题,将EM算法中E步的积分用Monte Carlo模拟来有效实现,使其适用性大大增强.但无论是EM算法,还是Monte Carlo EM算法,其收敛速度都是线性的,被缺损信息的倒数所控制,当缺损数据的比例很高时,收敛速度就非常缓慢.而Newton-Raphson算法在后验众数的附近具有二次收敛速率.本文提出Monte Carlo EM加速算法,将Monte Carlo EM算法与Newton-Raphson算法结合,既使得EM算法中的E步用Monte Carlo模拟得以实现,又证明了该算法在后验众数附近具有二次收敛速度.从而使其保留了Monte Carlo EM算法的优点,并改进了Monte Carlo EM算法的收敛速度.本文通过数值例子,将Monte Carlo EM加速算法的结果与EM算法、Monte Carlo EM算法的结果进行比较,进一步说明了Monte Carlo EM加速算法的优良性.
