Enhancing Computational Efficiency in State-Space Models Using Rao-Blackwellization and 2-Step Approximation

TL;DR

This paper introduces a Bayesian approach that combines Rao-Blackwellization and a 2-step approximation to improve computational efficiency in high-dimensional state-space models, reducing particle usage and processing time without sacrificing accuracy.

Contribution

The paper presents a novel combination of Rao-Blackwellization and a 2-step approximation for efficient Bayesian state-space modeling in high dimensions.

Findings

Significant reduction in particle count needed for accurate estimation.

Maintains high accuracy with reduced computational cost.

Effective in seasonal adjustment applications.

Abstract

This paper explores a Bayesian self-organization method for state-space models, enabling simultaneous state and parameter estimation without repeated likelihood calculations. While efficient for low-dimensional models, high-dimensional cases like seasonal adjustment require many particles. Using Rao-Blackwellization and a 2-step approximation, the method reduces particle use and computation time while maintaining accuracy, as shown in Monte Carlo evaluations.