Importance sampling for online variational learning
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910318708391936 |
|---|---|
| author | Chagneux, Mathis Gloaguen, Pierre Corff, Sylvain Le Olsson, Jimmy |
| author_facet | Chagneux, Mathis Gloaguen, Pierre Corff, Sylvain Le Olsson, Jimmy |
| contents | This article addresses online variational estimation in state-space models. We focus on learning the smoothing distribution, i.e. the joint distribution of the latent states given the observations, using a variational approach together with Monte Carlo importance sampling. We propose an efficient algorithm for computing the gradient of the evidence lower bound (ELBO) in the context of streaming data, where observations arrive sequentially. Our contributions include a computationally efficient online ELBO estimator, demonstrated performance in offline and true online settings, and adaptability for computing general expectations under joint smoothing distributions. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_02859 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Importance sampling for online variational learning Chagneux, Mathis Gloaguen, Pierre Corff, Sylvain Le Olsson, Jimmy Applications Methodology Machine Learning This article addresses online variational estimation in state-space models. We focus on learning the smoothing distribution, i.e. the joint distribution of the latent states given the observations, using a variational approach together with Monte Carlo importance sampling. We propose an efficient algorithm for computing the gradient of the evidence lower bound (ELBO) in the context of streaming data, where observations arrive sequentially. Our contributions include a computationally efficient online ELBO estimator, demonstrated performance in offline and true online settings, and adaptability for computing general expectations under joint smoothing distributions. |
| title | Importance sampling for online variational learning |
| topic | Applications Methodology Machine Learning |
| url | https://arxiv.org/abs/2402.02859 |