The predominant learning strategy for H(S)MMs is local search heuristics, of which the Baum-Welch/ expectation maximization (EM) algorithm is mostly used. It is an iterative learning procedure starting with a predefined topology and randomly-chosen initial parameters. However, state-of-the-art approaches based on arbitrarily defined state numbers and parameters can cause the risk of falling into a local optima and a low convergence speed with enormous number of iterations in learning which is computationally expensive. For models with persistent states, i.e. states with high self-transition probabilities, we propose a segmentation-based identification approach used as a pre-identification step to approximately estimate parameters based on segmentation and clustering techniques. The identified parameters serve as input of the Baum-Welch algorithm. Moreover, the proposed approach identifies automatically the state numbers. Experimental results conducted on both synthetic and real data show that the segmentation-based identification approach can identify H(S)MMs more accurately and faster than the current Baum-Welch algorithm.
Liu, T & Lemeire, J 2014, Effective and Efficient Identification of Persistent-state Hidden (semi-) Markov Models. in U Endriss & J Leite (eds), Proceedings of the 7th European Starting AI Researcher Symposium (STAIRS-2014) (Best Poster Award). vol. 264, IOS Press, 7th European Starting AI Researcher Symposium, STAIRS 2014, Prague, Czech Republic, 18/08/14.
Liu, T., & Lemeire, J. (2014). Effective and Efficient Identification of Persistent-state Hidden (semi-) Markov Models. In U. Endriss, & J. Leite (Eds.), Proceedings of the 7th European Starting AI Researcher Symposium (STAIRS-2014) (Best Poster Award) (Vol. 264). IOS Press.
@inproceedings{62bda9edfc3246c09a3da3783ec0f955,
title = "Effective and Efficient Identification of Persistent-state Hidden (semi-) Markov Models",
abstract = "The predominant learning strategy for H(S)MMs is local search heuristics, of which the Baum-Welch/ expectation maximization (EM) algorithm is mostly used. It is an iterative learning procedure starting with a predefined topology and randomly-chosen initial parameters. However, state-of-the-art approaches based on arbitrarily defined state numbers and parameters can cause the risk of falling into a local optima and a low convergence speed with enormous number of iterations in learning which is computationally expensive. For models with persistent states, i.e. states with high self-transition probabilities, we propose a segmentation-based identification approach used as a pre-identification step to approximately estimate parameters based on segmentation and clustering techniques. The identified parameters serve as input of the Baum-Welch algorithm. Moreover, the proposed approach identifies automatically the state numbers. Experimental results conducted on both synthetic and real data show that the segmentation-based identification approach can identify H(S)MMs more accurately and faster than the current Baum-Welch algorithm.",
keywords = "hidden Markov models (HMMs), hidden semi-Markov models (HSMMs),, Baum-Welch, local optima, model identification",
author = "Tingting Liu and Jan Lemeire",
note = "Endriss, U., Leite, J.; 7th European Starting AI Researcher Symposium, STAIRS 2014 ; Conference date: 18-08-2014 Through 22-08-2014",
year = "2014",
month = aug,
language = "English",
isbn = "978-1-61499-420-6",
volume = "264",
editor = "U. Endriss and J. Leite",
booktitle = "Proceedings of the 7th European Starting AI Researcher Symposium (STAIRS-2014) (Best Poster Award)",
publisher = "IOS Press",
address = "Netherlands",
}