kth.sePublications KTH
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Online Variational Sequential Monte Carlo
KTH, School of Engineering Sciences (SCI), Mathematics (Dept.).ORCID iD: 0000-0001-9380-1197
KTH, School of Engineering Sciences (SCI), Mathematics (Dept.).ORCID iD: 0000-0003-0772-846X
2024 (English)In: Proceedings of Machine Learning Research, ML Research Press , 2024, Vol. 235, p. 35039-35062Conference paper, Published paper (Refereed)
Abstract [en]

Being the most classical generative model for serial data, state-space models (SSM) are fundamental in AI and statistical machine learning. In SSM, any form of parameter learning or latent state inference typically involves the computation of complex latent-state posteriors. In this work, we build upon the variational sequential Monte Carlo (VSMC) method, which provides computationally efficient and accurate model parameter estimation and Bayesian latent-state inference by combining particle methods and variational inference. While standard VSMC operates in the offline mode, by re-processing repeatedly a given batch of data, we distribute the approximation of the gradient of the VSMC surrogate ELBO in time using stochastic approximation, allowing for online learning in the presence of streams of data. This results in an algorithm, online VSMC, that is capable of performing efficiently, entirely on-the-fly, both parameter estimation and particle proposal adaptation. In addition, we provide rigorous theoretical results describing the algorithm's convergence properties as the number of data tends to infinity as well as numerical illustrations of its excellent convergence properties and usefulness also in batch-processing settings.

Place, publisher, year, edition, pages
ML Research Press , 2024. Vol. 235, p. 35039-35062
Series
Proceedings of Machine Learning Research, ISSN 2640-3498
National Category
Computer Sciences Control Engineering
Identifiers
URN: urn:nbn:se:kth:diva-353953Scopus ID: 2-s2.0-85203797973OAI: oai:DiVA.org:kth-353953DiVA, id: diva2:1901029
Conference
41st International Conference on Machine Learning, ICML 2024, Vienna, Austria, Jul 21 2024 - Jul 27 2024
Note

QC 20241114

Available from: 2024-09-25 Created: 2024-09-25 Last updated: 2024-11-14Bibliographically approved
In thesis
1. Advances in Sequential Monte Carlo-based Statistical Learning: Online Algorithmic and Variational Inference
Open this publication in new window or tab >>Advances in Sequential Monte Carlo-based Statistical Learning: Online Algorithmic and Variational Inference
2024 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

This dissertation deals with sequential Monte Carlo (SMC) methods, also known as particle filters, focusing on online statistical learning in general state-space models (SSMs). SSMs are today an invaluable tool for modelling serial data in a variety of scientific and engineering disciplines such as automatic control, signal processing, biology, and finance. By including latent, Markovian states, SSMs offer high modeling flexibility, while SMC methods, which propagate recursively weighted samples using importance sampling and resampling techniques, are well-suited for state and model-parameter inference. This thesis, which consists of four papers, contributes to the enhancement of the online functionality of SMC, where 'online' refers to numerically stable estimation procedures with constant computational complexity and constant memory requirements over time.

The contribution of the thesis can be divided into two parts. The first part, which spans two papers, deals with algorithmic inference, focusing on both Bayesian state inference via SMC algorithms and the evaluation of their accuracy. Both papers address the challenges caused by the phenomenon of so-called particle-path degeneracy, an unavoidable issue caused by the resampling operation in SMC algorithms. Paper A presents AdaSmooth, a new algorithm developed to approximate expectations under joint-state posteriors in general SSMs when the objective is additive in the states. AdaSmooth, which uses adaptive backward-sampling techniques, avoids path-degeneracy and provides numerically stable estimates over infinite time horizons with reduced computational demands. Paper B presents ALVaR, an online method that consistently estimates the asymptotic variance of a particle filter based solely on the particle genealogy, avoiding extra sampling.

The second part combines SMC methods with variational inference and aims to develop online algorithms for parameter estimation in SSMs and proposal adaptation in particle filters, the latter being essential for accurate state posterior estimates. Paper C introduces OVSMC, which extends so-called variational SMC to online settings and allows simultaneous estimation of unknown model parameters and proposal kernel optimisation. Paper D proposes OSIWAE, which performs online variational inference by optimising a lower bound on the time-normalised limiting log-likelihood, resulting in a more theoretically grounded approach than OVSMC. OSIWAE ideally requires access to the filter state posteriors and their derivatives, which lack closed-form expressions in general. For this reason, a particle-based version, SMC-OSIWAE, which estimates the filter derivatives using AdaSmooth from Paper A, is developed.

Abstract [sv]

Denna avhandling handlar om sekventiella Monte Carlo-metoder (SMC), även kallade partikelfilter, med fokus på statistisk online-inlärning i generella tillståndsmodeller (TM). TM är idag ett ovärderligt verktyg för modellering av seriell data inom en rad olika vetenskapliga och tekniska fält såsom reglerteknik, signalbehandling, biologi och finans. Genom att inkludera latenta, markovska tillstånd erbjuder TM hög modelleringsflexibilitet, medan SMC-metoder, som propagerar viktade stickprov rekursivt med hjälp av vägd simulering och återsampling, är väl lämpade för tillståndsinferens och modellparameterskattning. Denna avhandling, som består av fyra artiklar, bidrar till att förbättra online-funktionaliteten hos SMC, där "online" avser numeriskt stabila skattningsalgoritmer med konstant beräkningskomplexitet och konstant minnesbehov över tid.

Avhandlingens bidrag kan delas in i två delar. Den första delen, vilken omfattar två artiklar, handlar om algoritmisk inferens, med fokus på både bayesiansk tillståndsinferens via SMC-metoder och utvärdering av deras noggrannhet. Båda artiklarna tar upp de utmaningar som orsakas av så kallad partikeltrajektoriedegenerering, ett oundvikligt problem som orsakas av återsamplingsoperationen i SMC-algoritmer. I Artikel A presenteras AdaSmooth, en ny algoritm som har utvecklats för att approximera väntevärden under den simultana posteriorifördelningen för tillstånden hos en generell TM när objektfunktionen är additiv i tillstånden. AdaSmooth, som använder adaptiva bakåtsamplingstekniker, undviker trajektoriedegenereringen och ger numeriskt stabila skattningar över oändliga tidshorisonter med låga beräkningskrav. Artikel B presenterar ALVaR, en online-metod som konsistent skattar den asymptotiska variansen hos ett partikelfilter baserat enbart på partikelgenealogin, utan ytterligare sampling. 

Den andra delen kombinerar SMC-metoder med variationsinferens och syftar till att utveckla online-algoritmer för modellparameterskattning i TM och adaption av förslagskärnan i partikelfilter, där det senare är avgörande för noggrannheten hos skattningen av tillståndsposteriorifördelningarna. Artikel C introducerar OVSMC, som utvidgar så kallad variations-SMC till online-situationen och möjliggör simultan skattning av okända modellparametrar och optimering av förslagskärnan. I Artikel D presenteras OSIWAE, som utför variationsinferens online genom att optimera en nedre gräns för den tidsnormaliserade asymptotiska log-likelihood-funktionen, vilket leder till ett mer teoretiskt välgrundat tillvägagångssätt än OVSMC. OSIWAE kräver tillgång till filterfördelningarna och deras derivator, som i allmänhet saknar slutna uttryck. Därför utvecklas en partikelbaserad version, SMC-OSIWAE, som uppskattar filterderivatorna med hjälp av AdaSmooth från Artikel A.

Place, publisher, year, edition, pages
Stockholm: KTH Royal Institute of Technology, 2024
Series
TRITA-SCI-FOU ; 2024:48
Keywords
sequential monte carlo, particle filters, state space models, additive smoothing, variance estimation, variational inference, sekventiella Monte Carlo-metoder, partikelfilter, tillståndsmodeller, algoritmisk inferens, variationsinferens
National Category
Probability Theory and Statistics
Research subject
Applied and Computational Mathematics; Applied and Computational Mathematics, Mathematical Statistics
Identifiers
urn:nbn:se:kth:diva-356328 (URN)978-91-8106-080-5 (ISBN)
Public defence
2024-12-13, F3 (Flodis), Lindstedtsvägen 26 & 28, Stockholm, 10:00 (English)
Opponent
Supervisors
Funder
Swedish Research Council, 2018-05230
Note

QC 2024-11-14

Available from: 2024-11-14 Created: 2024-11-13 Last updated: 2025-11-18Bibliographically approved

Open Access in DiVA

No full text in DiVA

Other links

ScopusPaper PDF

Authority records

Mastrototaro, AlessandroOlsson, Jimmy

Search in DiVA

By author/editor
Mastrototaro, AlessandroOlsson, Jimmy
By organisation
Mathematics (Dept.)
Computer SciencesControl Engineering

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 188 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf