Ensemble kalman variational objective: a variational inference framework for sequential variational auto-encoders
Open Access
- 1 January 2023
- journal article
- research article
- Published by Institute of Electronics, Information and Communications Engineers (IEICE) in Nonlinear Theory and Its Applications, IEICE
- Vol. 14 (4) , 691-717
Abstract
Time series model inference can be divided into modeling and optimization. Sequential VAEs have been studied as a modeling technique. As an optimization technique, methods combining variational inference (VI) and sequential Monte Carlo (SMC) have been proposed; however, they have two drawbacks: less particle diversity and biased gradient estimators. This paper proposes Ensemble Kalman Variational Objective (EnKO), a VI framework with the ensemble Kalman filter, to infer latent time-series models. Our proposed method efficiently learns the time-series models because of its particle diversity and unbiased gradient estimators. We demonstrate that our EnKO outperforms previous SMC-based VI methods in the predictive ability for several synthetic and real-world data sets.Keywords
This publication has 29 references indexed in Scilit:
- Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation2014
- Evaluating Methods to Account for System Errors in Ensemble Data AssimilationMonthly Weather Review, 2012
- TOPAZ4: an ocean-sea ice data assimilation system for the North Atlantic and ArcticOcean Science, 2012
- Implicit particle filters for data assimilationCommunications in Applied Mathematics and Computational Science, 2010
- Nonlinear data assimilation in geosciences: an extremely efficient particle filterQuarterly Journal of the Royal Meteorological Society, 2010
- Monte Carlo Smoothing for Nonlinear Time SeriesJournal of the American Statistical Association, 2004
- The Ensemble Kalman Filter: theoretical formulation and practical implementationOcean Dynamics, 2003
- Use of the breeding technique to estimate the structure of the analysis "errors of the day"Nonlinear Processes in Geophysics, 2003
- Long Short-Term MemoryNeural Computation, 1997
- Sequential data assimilation with a nonlinear quasi‐geostrophic model using Monte Carlo methods to forecast error statisticsJournal of Geophysical Research: Oceans, 1994