Nonparametric Learning of Switching Autoregressive Processes

Vector autoregressive (VAR) processes are useful in describing dynamical phenomena as diverse as speech, financial time-series, and the dancing of honey bees. However, such phenomena often exhibit