Ensemble Kalman Variational Objectives: Nonlinear Latent Trajectory Inference with A Hybrid of Variational Inference and Ensemble Kalman Filter

TL;DR

This paper introduces EnKO, a hybrid variational inference method combining ensemble Kalman filter techniques to improve nonlinear latent trajectory inference, addressing issues like particle degeneracy and biased gradients.

Contribution

EnKO is a novel hybrid approach that enhances latent state inference by leveraging ensemble Kalman filters within variational inference, improving efficiency and accuracy over existing SMC-based methods.

Findings

EnKO outperforms SMC methods in predictive accuracy.

EnKO demonstrates higher particle efficiency.

EnKO effectively infers nonlinear latent dynamics.

Abstract

Variational inference (VI) combined with Bayesian nonlinear filtering produces state-of-the-art results for latent time-series modeling. A body of recent work has focused on sequential Monte Carlo (SMC) and its variants, e.g., forward filtering backward simulation (FFBSi). Although these studies have succeeded, serious problems remain in particle degeneracy and biased gradient estimators. In this paper, we propose Ensemble Kalman Variational Objective (EnKO), a hybrid method of VI and the ensemble Kalman filter (EnKF), to infer state space models (SSMs). Our proposed method can efficiently identify latent dynamics because of its particle diversity and unbiased gradient estimators. We demonstrate that our EnKO outperforms SMC-based methods in terms of predictive ability and particle efficiency for three benchmark nonlinear system identification tasks.