Stochastic Systems Group
Home Research Group Members Programs  
Demos Calendar Publications Mission Statement Alumni

SSG Seminar Abstract


Switching Autoregressive Trees

Michael Siracusa
CSAIL, MIT


We consider the problem of inference over evolving dependency structures of multiple vector time-series. Specifically, we develop a fully Bayesian approach for inference over directed tree models as a means of examining the influence of time-series on each other. This analysis is embedded in a dynamic setting in which a latent variable indexes evolving tree structures. We demonstrate the utility of the method by analyzing the interaction of moving objects.



Problems with this site should be emailed to jonesb@mit.edu