Projects per year
Abstract
Particle smoothers are SMC (Sequential Monte Carlo) algorithms designed to approximate the joint distribution of the states given observations from a statespace model. We propose dSMC (deSequentialized Monte Carlo), a new particle smoother that is able to process T observations in O(log T) time on parallel architectures. This compares favorably with standard particle smoothers, the complexity of which is linear in T. We derive Lp convergence results for dSMC, with an explicit upper bound, polynomial in T. We then discuss how to reduce the variance of the smoothing estimates computed by dSMC by (i) designing good proposal distributions for sampling the particles at the initialization
of the algorithm, as well as by (ii) using lazy resampling to increase the number of particles used in dSMC. Finally, we design a particle Gibbs sampler based on dSMC, which is able to perform parameter inference in a statespace model at a O(log T) cost on parallel hardware.
of the algorithm, as well as by (ii) using lazy resampling to increase the number of particles used in dSMC. Finally, we design a particle Gibbs sampler based on dSMC, which is able to perform parameter inference in a statespace model at a O(log T) cost on parallel hardware.
Original language  English 

Number of pages  39 
Journal  Journal of Machine Learning Research 
Volume  23 
Publication status  Published  1 Aug 2022 
MoE publication type  A1 Journal articlerefereed 
Fingerprint
Dive into the research topics of 'DeSequentialized Monte Carlo: a parallelintime particle smoother'. Together they form a unique fingerprint.Projects
 1 Finished

Parallel and distributed computing for Bayesian graphical models
Särkkä, S., Emzir, M., Corenflos, A., Hassan, S. S., Ma, X., Merkatas, C., Yaghoobi, F. & Yamin, A.
04/09/2019 → 31/12/2022
Project: Academy of Finland: Other research funding