Please use this identifier to cite or link to this item:
|Title:||A hidden Markov model with dependence jumps for predictive modeling of multidimensional time-series||Authors:||Petropoulos, Anastasios
Chatzis, Sotirios P.
|Keywords:||Dependence jumps;Variable order;Expectation-maximization;Hidden Markov models;Temporal dynamics||Category:||Computer and Information Sciences||Field:||Natural Sciences||Issue Date:||1-Oct-2017||Publisher:||ELSEVIER SCIENCE INC||Source:||INFORMATION SCIENCES, Volume: 412, Pages: 50-66, Published: OCT 2017||metadata.dc.doi:||http://dx.doi.org/10.1016/j.ins.2017.05.038||Journal:||Information Sciences||Abstract:||Hidden Markov models (HMMs) are a popular approach for modeling sequential data, typically based on the assumption of a first- or moderate-order Markov chain. However, in many real-world scenarios the modeled data entail temporal dynamics the patterns of which change over time. In this paper, we address this problem by proposing a novel HMM formulation, treating temporal dependencies as latent variables over which inference is performed. Specifically, we introduce a hierarchical graphical model comprising two hidden layers: on the first layer, we postulate a chain of latent observation-emitting states, the temporal dependencies between which may change over time; on the second layer, we postulate a latent first-order Markov chain modeling the evolution of temporal dynamics (dependence jumps) pertaining to the first-layer latent process. As a result of this construction, our method allows for effectively modeling non-homogeneous observed data, where the patterns of the entailed temporal dynamics may change over time. We devise efficient training and inference algorithms for our model, following the expectation-maximization paradigm. We demonstrate the efficacy and usefulness of our approach considering several real-world datasets.||URI:||http://ktisis.cut.ac.cy/handle/10488/10515||ISSN:||0020-0255||Rights:||(C) 2017 Elsevier Inc. All rights reserved.||Type:||Article|
|Appears in Collections:||Άρθρα/Articles|
Show full item record
checked on Dec 13, 2017
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.