A Global Optimal Gaussian Mixture Reduction Approach Based on Integer Linear Programming  

作　　者：ZHU Hongyan ZHAI Qiaozhu 

机构地区：[1]School of Electronic and Information Engineering, Xi'an Jiaotong University, Xi'an 710049, China [2]Ministry of Education Key Laboratory for Intelligent Networks and Network Security, Xi'an Jiaotong University, Xi'an 710049, China

出　　处：《Chinese Journal of Electronics》2013年第4期763-768,共6页电子学报（英文版）

基　　金：This work is supported by the National Natural Science Foundation of China （No.61203220, No.61174146）, the Program for New Century Talents of Education Ministry （No.NCET-08-0432）, and the Foundation for Authors of National Outstanding Doctoral Dissertation （No.201047）.

摘　　要：In many applications, the Gaussian mix- ture serves as an important probabilistic representation of the system state. A global optimal Gaussian mixture re- duction （GMR） approach based on Integer linear program- ming （ILP） is developed in this paper. Firstly, a Gaussian base set is constructed with partial merging of components of the original mixture. Secondly, by introducing auxiliary variables reasonably, the original problem of selecting the best candidates from the given Gaussian base set is formu- lated as an ILP problem. Finally, a global optimal solution to GMR is obtained by solving the ILP problem. The global optimum property enables it as a basis for perfor- mance comparison with different GMR algorithms.

关 键 词：Gaussian mixture reduction Integer lin- ear programming Component merging Integral squared difference Global optimal solution. 

分 类 号：O221.4[理学—运筹学与控制论] TN911.7[理学—数学]