Bayesian Tracking within a Feedback Sensing Environment: Estimating Interacting, Spatially Constrained Complex Dynamical Systems from Multiple Sources of Controllable Devices.

摘要

This grant led to developments in flexible models for complex time series in a range of applications with a focus on Bayesian and Bayesian nonparametric methods. Three fundamental challenges were tackled: (i) capturing evolving correlations in high-dimensional time series with possible missing or irregularly-spaced observations, (ii) performing diverse subset selection over time, and (iii) automatically learning an unknown set of simple underlying temporal structures to describe complex dynamical phenomena. Each of these methods was applied in a range of application domains including neuroimaging, diverse document selection, speaker diarization, stock modeling, and target tracking.