We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Particle Smoothing for Hidden Diffusion Processes: Adaptive Path Integral Smoother

Abstract: Particle smoothing methods are used for inference of stochastic processes based on noisy observations. Typically, the estimation of the marginal posterior distribution given all observations is cumbersome and computational intensive. In this paper, we propose a simple algorithm based on path integral control theory to estimate the smoothing distribution of continuous-time diffusion processes with partial observations. In particular, we use an adaptive importance sampling method to improve the effective sampling size of the posterior over processes given the observations and the reliability of the estimation of the marginals. This is achieved by estimating a feedback controller to sample efficiently from the joint smoothing distributions. We compare the results with estimations obtained from the standard Forward Filter/Backward Simulator for two diffusion processes of different complexity. We show that the proposed method gives more reliable estimations than the standard FFBSi when the smoothing distribution is poorly represented by the filter distribution.
Comments: 16 pages, 13 figures
Subjects: Machine Learning (cs.LG); Computation (stat.CO)
DOI: 10.1109/TSP.2017.2686340
Cite as: arXiv:1605.00278 [cs.LG]
  (or arXiv:1605.00278v2 [cs.LG] for this version)

Submission history

From: Hans-Christian Ruiz Dipl-Phys [view email]
[v1] Sun, 1 May 2016 16:56:49 GMT (1408kb,D)
[v2] Mon, 6 Mar 2017 15:15:00 GMT (1771kb,D)

Link back to: arXiv, form interface, contact.