期刊名称：IAENG International Journal of Computer Science
印刷版ISSN：1819-656X
电子版ISSN：1819-9224
出版年度：2020
卷号：47
期号：4
语种：English
出版社：IAENG - International Association of Engineers
摘要：Hidden Markov models (HMMs) are applied to many problems of computational Molecular Biology. In a predictive task, the HMM is endowed with a decoding algorithm in order to assign the most probable path of states, and in turn the class labelling, to an unknown sequence. In this paper we introduce a novel decoding algorithm Log-posterior-best (LPB) which combines the log-odd posterior probability and 1-best algorithms. LPB is a two steps process: first the Log odd probability of each state is computed and then the best allowed label path through the model is evaluated by a 1-best algorithm. We show that our LPB decoding performs better than other existing algorithms in some computational biological problems such as gene finding in prokaryotes.
