Personal tools
You are here: Home Events ANC Workshop: Botond Cseke, Chair: Guido Sanguinetti

ANC Workshop: Botond Cseke, Chair: Guido Sanguinetti

— filed under:

  • ANC Workshop Talk
When Oct 01, 2013
from 11:00 AM to 12:00 PM
Where IF 4.31/4.33
Add event to calendar vCal

Approximate inference in latent Gaussian (OU) processes from continuous time observations

We apply expectation constraints based approximate inference in latent Gaussian continuous time stochastic dynamical systems observed with both discrete and continuous time with noise (OU-process prior, continuous and discrete time likelihood terms/factors).  We show that the continuous time limit exists and results in a neat, hybrid fixed point iteration: (1) expectation propagation updates for discrete time terms and (2) variational updates for the continuous time updates. We also introduce post-inference corrections methods that improve on the marginals of the approximation. This approach extends the classical Kalman-Bucy smoothing procedure to non-Gaussian observations, enabling continuous-time inference in a variety of models, including spiking neuronal models (state-space models with point process observations) and box likelihood models. Experimental results on real and simulated data demonstrate relatively high distributional accuracy and significant computational savings compared to discrete-time approaches in a neural application.