Fuzzy Hidden Markov Model And Its Application In Biological Sequence Alignment

By comparing biological sequences,we can predict the function of unknown sequences.At present,the most cutting-edge sequence alignment algorithms include dynamic programming algorithm,genetic algorithm,artificial neural network and Hidden Markov Model.This paper is based on the Hidden Markov Model to study the biological sequence alignment.The Hidden Markov Model needs to study three basic problems:valuation problem,decoding problem and parameter estimation.For each problem,we have a corresponding solution.The Hidden Markov Model assumes that the observed values are independent,which is not the case in practice.In this paper,the independent hypothesis of Hidden Markov Model is relaxed by introducing fuzzy measure and fuzzy integral,and then the biological sequence comparison is carried out.After introducing the fuzzy theory,we call the Hidden Markov Model as fuzzy Hidden Markov Model.In this paper,we introduce the Hidden Markov Model,and propose three problems to be solved by the Hidden Markov Model.Then the application and limitation of the Hidden Markov Model in biological sequence alignment are introduced.In view of the limitations of the Hidden Markov Model,we propose an improved model-fuzzy Hidden Markov Model for biological sequence alignment.This model introduces fuzzy theory on the basis of the Hidden Markov Model.It is known by statistical knowledge that the measure has additivity,and the main characteristic of fuzzy measure is nonadditivity.Because the fuzzy measure is monotonous,the statistical independence of the Hidden Markov Model is relaxed after the fuzzy measure is introduced.In this paper,the algorithm of fuzzy Hidden Markov Model is presented:fuzzy forward algorithm,fuzzy backward algorithm,fuzzy Viterbi algorithm and fuzzy EM algorithm.If we introduce the probability measure and the Choquet fuzzy integral,the fuzzy Hidden Markov Model becomes the classic Hidden Markov Model.The fuzzy measure includes probability measure,likelihood measure,? fuzzy measure,trust measure,necessity measure and self-dual measure.,etc.In this paper we introduce the probability measure(the most conservative fuzzy measure)and the Choquet fuzzy integral.The model of probability measure and Choquet fuzzy integral is applied in the experiment.After the empirical analysis,the improved fuzzy Hidden Markov Model has better effect on the biological sequence comparison.