Backward Smoothing versus Fixed-Lag Smoothing in Particle Filters

2026-02-14

AI Summary

Particle smoothing in nonlinear, non-Gaussian state-space models is computationally expensive, limiting its practical applicability. This study systematically compares backward smoothers (FFBS/FFBSm) with fixed-lag smoothing in terms of the trade-off between accuracy and computational efficiency, and evaluates O(m) approximations based on subsampling and local neighborhood constraints. Experimental results reveal performance differences across noise distributions: under a fixed particle budget, FFBS/FFBSm achieve higher accuracy, yet fixed-lag smoothing typically outperforms them when constrained by equal computation time. Furthermore, while the FFBSm approximation performs well under Gaussian dynamics, its advantage diminishes markedly under heavy-tailed transition distributions, thereby clarifying its practical limitations and scope of applicability.

Technology Category

Intelligent Robots: State EstimationSearch and Optimization: Sampling/Simulation-based SearchReasoning under Uncertainty: Stochastic Optimization

Application Category

Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphsUser Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systemsWeb Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Particle smoothing enables state estimation in nonlinear and non-Gaussian state-space models, but its practical use is often limited by high computational cost. Backward smoothing methods such as the Forward Filter Backward Smoother (FFBS) and its marginal form (FFBSm) can achieve high accuracy, yet typically require quadratic computational complexity in the number of particles. This paper examines the accuracy--computational cost trade-offs of particle smoothing methods through a trend-estimation example. Fixed-lag smoothing, FFBS, and FFBSm are compared under Gaussian and heavy-tailed (Cauchy-type) system noise, with particular attention to O(m) approximations of FFBSm based on subsampling and local neighborhood restrictions. The results show that FFBS and FFBSm outperform fixed-lag smoothing at a fixed particle number, while fixed-lag smoothing often achieves higher accuracy under equal computational time. Moreover, efficient FFBSm approximations are effective for Gaussian transitions but become less advantageous for heavy-tailed dynamics.

Problem

• particle smoothing
• computational cost
• accuracy trade-off
• non-Gaussian models
• fixed-lag smoothing

Innovation

• particle smoothing
• fixed-lag smoothing
• FFBSm approximation
• computational complexity
• heavy-tailed noise