Explicit-Duration Markov Switching Models
By Silvia Chiappa. 2014
23 citations recorded by the source corpus.
Markov switching models MSMs are probabilistic models that employmultiple sets of parameters to describe different dynamic regimesthat a time series may exhibit at different periods of time. Theswitching mechanism between regimes is controlled by unobserved randomvariables that form a first-order Markov chain. Explicit-durationMSMs contain additional variables that explicitly model the distributionof time spent in each regime. This allows to define duration distributionsof any form, but also to impose complex dependence betweenthe observations and to reset the dynamics to initial conditions. Modelsthat focus on the first two properties are most commonly known as hiddensemi-Markov models or segment models, whilst models that focuson the third property are most commonly known as changepoint modelsor reset models. In this monograph, we provide a description of explicitdurationmodelling by categorizing the different approaches into threegroups, which differ in encoding in the explicit-duration variables differentinformation about regime change/reset boundaries. The approachesare described using the formalism of graphical models, which allows tographically represent and assess statistical dependence and thereforeto easily describe the structure of complex models and derive inferenceroutines. The presentation is intended to be pedagogical, focusingon providing a characterization of the three groups in terms of modelstructure constraints and inference properties. The monograph is supplementedwith a software package that contains most of the modelsand examples described. The material presented should be useful toboth researchers wishing to learn about these models and researcherswishing to develop them further.The supplementary data for this issue will be available soon.
Subfields: Artificial Intelligence, Control and Systems Engineering
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
In Control and Systems Engineering
Citations in the corpus
- Path-Specific Counterfactual Fairness (2018, methodology): “Causal relationships among random variables can be visually expressed using Graphical Causal Models (GCMs), which are a special case of Graphical Models (see Chiappa (2014) for a quick introduction on GMs).”