Conditional Measurement Density Estimation in Sequential Monte Carlo via Normalizing Flow
This work addresses a specific problem in sequential Monte Carlo for real-world applications, but it is incremental as it builds on existing differentiable particle filter frameworks.
The paper tackled the challenge of tuning measurement models in sequential Monte Carlo methods by proposing conditional normalizing flows to learn valid probability densities, resulting in improved estimation performance and faster training convergence in a visual tracking experiment.
Tuning of measurement models is challenging in real-world applications of sequential Monte Carlo methods. Recent advances in differentiable particle filters have led to various efforts to learn measurement models through neural networks. But existing approaches in the differentiable particle filter framework do not admit valid probability densities in constructing measurement models, leading to incorrect quantification of the measurement uncertainty given state information. We propose to learn expressive and valid probability densities in measurement models through conditional normalizing flows, to capture the complex likelihood of measurements given states. We show that the proposed approach leads to improved estimation performance and faster training convergence in a visual tracking experiment.