Variational Resampling

Oskar Kviman, Víctor Branchini Nicolaand Elvira, Jens Lagergren
Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, PMLR 238:3286-3294, 2024.

Abstract

We cast the resampling step in particle filters (PFs) as a variational inference problem, resulting in a new class of resampling schemes: variational resampling. Variational resampling is flexible as it allows for choices of 1) divergence to minimize, 2) target distribution to input to the divergence, and 3) divergence minimization algorithm. With this novel application of VI to particle filters, variational resampling further unifies these two powerful and popular methodologies. We construct two variational resamplers that replicate particles in order to maximize lower bounds with respect to two different target measures. We benchmark our variational resamplers on challenging smoothing tasks, outperforming PFs that implement the state-of-the-art resampling schemes.

Cite this Paper

BibTeX
@InProceedings{pmlr-v238-kviman24a, title = { Variational Resampling }, author = {Kviman, Oskar and Branchini, Nicolaand Elvira, V\'{i}ctor and Lagergren, Jens}, booktitle = {Proceedings of The 27th International Conference on Artificial Intelligence and Statistics}, pages = {3286--3294}, year = {2024}, editor = {Dasgupta, Sanjoy and Mandt, Stephan and Li, Yingzhen}, volume = {238}, series = {Proceedings of Machine Learning Research}, month = {02--04 May}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v238/kviman24a/kviman24a.pdf}, url = {https://proceedings.mlr.press/v238/kviman24a.html}, abstract = { We cast the resampling step in particle filters (PFs) as a variational inference problem, resulting in a new class of resampling schemes: variational resampling. Variational resampling is flexible as it allows for choices of 1) divergence to minimize, 2) target distribution to input to the divergence, and 3) divergence minimization algorithm. With this novel application of VI to particle filters, variational resampling further unifies these two powerful and popular methodologies. We construct two variational resamplers that replicate particles in order to maximize lower bounds with respect to two different target measures. We benchmark our variational resamplers on challenging smoothing tasks, outperforming PFs that implement the state-of-the-art resampling schemes. } }
Endnote
%0 Conference Paper %T Variational Resampling %A Oskar Kviman %A Víctor Branchini, Nicolaand Elvira %A Jens Lagergren %B Proceedings of The 27th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2024 %E Sanjoy Dasgupta %E Stephan Mandt %E Yingzhen Li %F pmlr-v238-kviman24a %I PMLR %P 3286--3294 %U https://proceedings.mlr.press/v238/kviman24a.html %V 238 %X We cast the resampling step in particle filters (PFs) as a variational inference problem, resulting in a new class of resampling schemes: variational resampling. Variational resampling is flexible as it allows for choices of 1) divergence to minimize, 2) target distribution to input to the divergence, and 3) divergence minimization algorithm. With this novel application of VI to particle filters, variational resampling further unifies these two powerful and popular methodologies. We construct two variational resamplers that replicate particles in order to maximize lower bounds with respect to two different target measures. We benchmark our variational resamplers on challenging smoothing tasks, outperforming PFs that implement the state-of-the-art resampling schemes.
APA
Kviman, O., Branchini, Nicolaand Elvira, V. & Lagergren, J.. (2024). Variational Resampling . Proceedings of The 27th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 238:3286-3294 Available from https://proceedings.mlr.press/v238/kviman24a.html.

Related Material