Restrained Resampling for Sequential Monte Carlo Methods

Inpyo Hong, Daeyoung Kim, and Daehong Kim


Resampling, Sequential Monte Carlo Methods, Particle Filters, Parameter Estimation, Bayesian Estimation


Sequential Monte Carlo methods rely on resampling. Even though powerful resampling schemes now exist, they still have two drawbacks. First, they remove a large number of samples so that the diversity of samples becomes reduced. Second, they lose large weight information so that the error of approximation becomes high. As a result, sequential Monte Carlo methods suffer from low sampling efficiency. So, this paper aims to improve the methods by proposing a new resampling scheme, Restrained Resampling, designed to overcome the two drawbacks. Our proposed scheme employs a deterministic approach to decrease the number of removed samples and keeps weights such that the loss of weight information is minimized. By doing so, our scheme increases the diversity of samples and approximation performance. These improve the sampling efficiency of the methods. We will show these results through extensive analyses and two recursive Bayesian estimation examples.

Important Links:

Go Back