Juhuslike jadade võrdlemine kolmekaupa Markovi mudeli korral
Laen...
Kuupäev
Autorid
Ajakirja pealkiri
Ajakirja ISSN
Köite pealkiri
Kirjastaja
Tartu Ülikool
Abstrakt
Töös kirjeldatakse ja uuritakse kahe juhusliku jada võrdlemist kolmekaupa Markovi mudeli korral. Alguses tutvustatakse kolmekaupa Markovi mudelit ja pikima ühisjada pikkuse sarnasusmõõtu. Seejärel kirjeldatakse varjatud Markovi mudeli (VMM) erijuhtu kolmekaupa Markovi mudelist. Viimases peatükis uuritakse varjatud Markovi mudeleid, kus varjatud protsess on juhuslik ekslemine täisarvudel. Töö käigus läbi viidud simulatsioonid näitasid, et positiivselt korreleeritud emissioonide saartega mudeli ning positiivselt ja negatiivselt korreleeritud emissioonidega seisundites eksleva mudeli korral on pikima ühisjada pikkus suurem kui iid-jadade korral.
In this thesis, we describe and explore comparison of two random sequences in the case of a triplet Markov model. At first, we describe the triplet Markov model and the length longest common subsequence (LCS) similarity measurement. Then we explore the hidden Markov model (HMM) as a special case of the triplet Markov model. Finally, we explore HMM-s where the hidden process is a random walk on the integer line. The simulations conducted during this study demonstrated that for both the model with islands of positively correlated emissions and the model randomly walking between states of positively and negatively correlated emissions, the length of the longest common subsequence is greater than for i.i.d sequences.
In this thesis, we describe and explore comparison of two random sequences in the case of a triplet Markov model. At first, we describe the triplet Markov model and the length longest common subsequence (LCS) similarity measurement. Then we explore the hidden Markov model (HMM) as a special case of the triplet Markov model. Finally, we explore HMM-s where the hidden process is a random walk on the integer line. The simulations conducted during this study demonstrated that for both the model with islands of positively correlated emissions and the model randomly walking between states of positively and negatively correlated emissions, the length of the longest common subsequence is greater than for i.i.d sequences.
Kirjeldus
Märksõnad
juhuslikud protsessid, Markovi ahelad, jadad, stochastic processes, Markov chains, sequences