Differentiable Systematic Resampling for Variational Sequential Monte Carlo
What it is
This paper introduces Differentiable Systematic Resampling (DSR), a temperature-controlled relaxation that makes the discrete resampling step in particle filters differentiable. It preserves the structure of systematic resampling while enabling gradient flow, allowing for end-to-end training in variational sequential Monte Carlo.
The problem it solves
The discrete nature of standard resampling in particle filters prevents the use of gradient-based learning methods in variational sequential Monte Carlo.
What you could use it for
- Training variational filters for nonlinear state estimation
- Learning dynamics in stochastic dynamical systems
- Processing real-world time-series data like handwriting
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.