马尔可夫信源的熵分类法

Classifying method employing the entropy of a Markov chain

在对一组具有相关关系的随机序列数据进行分类时,若序列很长,将这组数据看成是马尔可夫链的一个样本,于是,该问题就成为判别这组数据源于哪个马尔可夫信源总体的问题。在介绍马尔可夫信源熵及其极限熵的基础上,提出三个判别准则,并以DNA序列分类为例,分别进行验证,结果表明,判别准则3的性能优于其他方法。

When there exists a necessity of classifying a correlated random sequence which shows a greater number of the sample points, this sequence tends to be regarded as a sample of Markov chain. Then, there arises the problem of distinguishing which Markov information source this sequence comes from. According to the basic principles of Markov information source's entropy and limit entropy, this paper presents three diagnostic regulations. Applying these diagnostic regulations to the DNA sequence's classification shows that the 3rd diagnostic regulation gives the best performance.