摘要
EM算法的计算强度较大,且当数据集较大时,计算效率较低。为此,提出了基于部分E步的混合EM算法,降低了算法的计算强度,提高了算法对数据集大小的适应能力,并且保持了EM算法的收敛特性。最后通过将算法应用于大的数据集,验证了该算法能减少计算强度。
EM algorithm often needs great computational costs. And its computing is inefficient when the data sets are large. A mixed EM algorithm based on partial E-steps method was presented which can reduce the intensity of computation, make it adapted to the scale of data sets better and have the standard convergence guarantee of EM. It is verified that the mixed EM algorithm can reduce computational costs evidently through its application to large data sets.
出处
《计算机应用》
CSCD
北大核心
2006年第8期1884-1887,共4页
journal of Computer Applications
基金
国家自然科学基金资助项目(60475040)
河南省自然科学基金资助项目(0511012200)
河南省高校青年骨干教师资助计划项目(2002261)
