CAJ | 학술논문

本文参照直线上隐Markov模型的概念，给出有限树指标隐Markov链的定义。在该定义中，树指标隐Markov链由两个树指标随机过程组成，其中第一个树指标随机过程是树指标Markov链，是不能被直接观测到的隐藏链；第二个树指标随机过程是可被观测的且关于第一个树指标随机过程条件独立，对于树上的任意一个顶点，第二个随机过程此处的取值只信赖于隐藏链中此处的取值。最后，我们给出了树指标隐Markov链的三个等价定义。
본문삼조직선상은Markov모형적개념，급출유한수지표은Markov련적정의。재해정의중，수지표은Markov련유량개수지표수궤과정조성，기중제일개수지표수궤과정시수지표Markov련，시불능피직접관측도적은장련；제이개수지표수궤과정시가피관측적차관우제일개수지표수궤과정조건독립，대우수상적임의일개정점，제이개수궤과정차처적취치지신뢰우은장련중차처적취치。최후，아문급출료수지표은Markov련적삼개등개정의。
In this paper, we give the definition of tree indexed hidden Markov chain with finite state space based on the concept of hidden Markov model. In our definition, tree indexed hidden Markov chain consists of two tree indexed random processes. The underlying process is a tree indexed Markov chain and can not be observed, and the second process is conditional independent of the former. For the arbitrary vertex in tree, the second process only dependents on the underlying process. Finally, we propose three equivalent definitions.