State duration and interval modeling in hidden semi-Markov model for sequential data analysis

Abstract

Sequential data modeling and analysis have become indispensable tools for analyzing sequential data such as time-series data because a larger amount of sensed event data have become available. These methods capture the sequential structure of data of interest, such as inputoutput relationship and correlation among datasets. However, since most studies in this area are specialized or limited for their respective applications, rigorous requirement analysis on such a model has not been examined in a general point of view. Hence, we particularly examine the structure of sequential data, and extract the necessity of "state duration" and "state duration" of events for efficient and rich representation of sequential data. Specifically addressing the hidden semi-Markov model (HSMM) that represents such state duration inside a model, we attempt to newly add representational capability of state interval of events onto HSMM. To this end, we propose two extended models; one is interval state hidden semi-Markov model (IS-HSMM) to express the length of state interval with a special state node designated as "interval state node". The other is interval length probability hidden semi-Markov model (ILP-HSMM) which represents the length of state interval with a new probabilistic parameter "interval length probability." From exhaustive simulations, we show superior performances of the proposed models in comparison with HSMM. To the best of our knowledge, our proposed models are the first extensions of HMM to support state interval representation as well as state duration representation.

A Hybrid RNN-HMM Approach for Weakly Supervised Temporal Action Segmentation

“…in [41]. Since then it has been used in different contexts such as general modeling in case of explicit state duration HMMs [42] but also for speech synthesis [43] or the generation and decoding of temporal sequences, such as music [44].…”

Multivariate time series analysis from a Bayesian machine learning perspective

“…Our proposed methodology of feature selection on the regression component of time series can naturally extend the univariate analysis to multivariate analysis, extend the dimension of predictors by means of feature selection, and extend the corresponding analysis to the time-dependent case which is more realistic in many situations. Other extensions regarding methodology improvement may be seen from the following perspectives, such as adding on spatial analysis to generate a spatio-temporal framework (see [2]), boosting conditional probability estimators (see [15]), generalizing to universal probability-free prediction (see [38]), extending linear regression to nonlinear regression (see [22]), extending state space model (also named hidden Markov model) to hidden semi-Markov model (see [28]).…”

Consistency-aware and Inconsistency-aware Graph-based Multi-view Clustering

“…Many machine learning applications such as image classification, social networks, chemistry, signal processing, web analysis, item recommendation, and bioinformatics analysis usually exhibit some structured data, which might include trees, groups, clusters, paths, sequences [2,3,4], and graphs [5]. Recent advances in information-retrieval technologies enable collection of such structured data with heterogeneous features from multi-view data.…”

“…Source code available at https://github.com/frash1989/ELM-MVClustering/tree/master/RMSC-ELM 3. Source code available at http://legacydirs.umiacs.umd.edu/ ~abhishek/code_coregspectral.zip 4.…”