Differentiable interacting multiple model particle filtering

John-Joseph Brady

doi:10.1016/j.sigpro.2025.110166

Differentiable interacting multiple model particle filtering

2025-06-30

0

PRE

AI

DOI：10.1016/j.sigpro.2025.110166

Original

Original request for help

Save

Abstract

En 中文

• This paper addresses the issue of parameter learning in state-space systems with behaviour jumps (or regime switching). • We propose the first filter that can learn the switching process and underlying models simultaneously. • The new filter performs better than the previous (non-filtering) state-of-the-art.

Keywords:

62M20

62F12

Sequential Monte Carlo

Differentiable particle filtering

Regime switching

Differentiable interacting multiple model particle filtering

Abstract

Journal

Organization