Deep parameterizations of pairwise and triplet Markov models for unsupervised classification of sequential data

Non-stationary data segmentation with hidden evidential semi-Markov chains

“…However, when each 𝑈 𝑛 take its values in a finite set, MPM is still computable in a similar way it is done in HMCs and PMCs. TMCs have been successively used in image segmentation [24], [25], [26], normalized difference vegetation index modelling [27], activity classification [28], repayment of consumer loan modelling [29], non-stationary fuzzy data segmentation [30], [31], or still respiratory signal analysis [32]. As 𝑈 𝑁 is arbitrary in the general case, it may not have a practical interpretation.…”