Variational Inference for Diffusion Processes
There is a more recent version of this eprint available. Click here to view it.
Diffusion processes are a family of continuous-time continuous-state stochastic processes that are in general only partially observed. The joint estimation of the forcing parameters and the system noise (volatility) in these dynamical systems is a crucial, but non-trivial task, especially when the system is nonlinear and multi-modal. We propose a variational treatment of diffusion processes, which allows us to compute type II maximum likelihood estimates of the parameters by simple gradient techniques and which is computationally less demanding than most MCMC approaches. We also show how a cheap estimate of the posterior over the parameters can be constructed based on the variational free energy.Experiments show that our variational approximation is viable and that the results are very promising as the variational approximation outperforms standard Gaussian process regression for non-Gaussian Markov processes.
Available Versions of this Item