基于信息扩散的极大似然估计

Maximum Likelihood Estimation Based on the Principle of Information Diffusion

摘要: 提出了扩散极大似然估计方法,利用实际观测值的概率密度函数的信息扩散估计,代替了对观测值分布的主观假设,从而具有很强的自适应性。最后设计了两个算例,说明了扩散极大似然估计的过程,并考察了扩散极大似然估计的特性。

Abstract: This paper introduces the principle of information diffusion and information diffusion estimation(IDE).And with IDE,the observation distribution can be estimated easily.Once the observation distribution is determined,the parameters can be estimated with the maximum likelihood.With IDE as the basis,the diffusion maximum likelihood(DML) estimation is presented.DML is supposed to enjoy priority because of its being free from any supposition of observation distribution.With two simulative persuasive examples,the high self-adapting and robustness of DML are discussed.