RECORD
Scalable Bayesian Learning for State Space Models using Variational Inference with SMC Samplers
Machine Learning · Submitted 23 May 2018
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim We build up on recent advances in computational statistics that combine variational methods with sequential Monte Carlo sampling and we demonstrate the advantages of performing full Bayesian inference over the static parameters rather than just performing variational EM approximations.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
We present a scalable approach to performing approximate fully Bayesian inference in generic state space models. The proposed method is an alternative to particle MCMC that provides fully Bayesian inference of both the dynamic latent states and the static parameters of the model. We build up on recent advances in computational statistics that combine variational methods with sequential Monte Carlo sampling and we demonstrate the advantages of performing full Bayesian inference over the static parameters rather than just performing variational EM approximations. We illustrate how our approach enables scalable inference in multivariate stochastic volatility models and self-exciting point process models that allow for flexible dynamics in the latent intensity function.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | We present a scalable approach to performing approximate fully Bayesian inference in generic state space models. | contribution 1.00 | 0.89 | 0.26 | asserted 1.00 |
| s1 | The proposed method is an alternative to particle MCMC that provides fully Bayesian inference of both the dynamic latent states and the static parameters of the model. | contribution 0.73 | 0.68 | 0.21 | asserted 1.00 |
| s2 | We build up on recent advances in computational statistics that combine variational methods with sequential Monte Carlo sampling and we demonstrate the advantages of performing full Bayesian inference over the static parameters rather than just performing variational EM approximations. | method 0.44 | 0.64 | 0.28 | asserted 1.00 |
| s3 | We illustrate how our approach enables scalable inference in multivariate stochastic volatility models and self-exciting point process models that allow for flexible dynamics in the latent intensity function. | contribution 0.54 | 0.89 | 0.17 | asserted 0.99 |
What was examined, and what was not
- arXiv listing abstract v3
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
We present a scalable approach to performing approximate fully Bayesian inference in generic state space models. The proposed method is an alternative to particle MCMC that provides fully Bayesian inference of both the dynamic latent states and the static parameters of the model. We build up on recent advances in computational statistics that combine variational methods with sequential Monte Carlo sampling and we demonstrate the advantages of performing full Bayesian inference over the static parameters rather than just performing variational EM approximations. We illustrate how our approach enables scalable inference in multivariate stochastic volatility models and self-exciting point process models that allow for flexible dynamics in the latent intensity function.
Human review
How the literature uses it
What this paper does with the arXiv papers it cites
What it cites · 4 sentences without a confident label
Uses the method or tool of 1 paper
-
We take Monte Carlo samples of the expectation above and optimize the bound using Adam [ref] .
Adam: A Method for Stochastic Optimization · Optimization of the variational bound · 1.00
Uses data or values from 1 paper
-
The data is available as part of the tick library [ref] with $4$ event types: (i) mid-price up moves, (ii) mid-price down moves, (iii) buyer-initiated trades leaving the mid-price unchanged and (iv) seller-initiated trades not changing the mid.
Tick: a Python library for statistical learning, with a particular emphasis on time-dependent modelling · Non-linear stochastic Hawkes processes · 0.85
Compares with 1 paper
-
The representation in Corollary [ref] allows us to contrast the variational bound to previously considered sequential VAE frameworks [ref] .
Black box variational inference for state space models · Related Work · 0.87
Cites as background 1 paper
-
Recent work [ref] have advocated the use of neural networks in a black-box treatment of point process dynamics.
Joint Modeling of Event Sequence and Time Series with Attentional Twin Recurrent Neural Networks · Non-linear stochastic Hawkes processes · 0.94
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.