Variational Marginal Particle Filters
This work addresses a specific bottleneck in state space model inference, offering incremental improvements for researchers in probabilistic modeling.
The authors tackled the challenge of variational inference for state space models by proposing the variational marginal particle filter (VMPF), a differentiable and reparameterizable objective based on an unbiased estimator, which yields tighter bounds than previous methods.
Variational inference for state space models (SSMs) is known to be hard in general. Recent works focus on deriving variational objectives for SSMs from unbiased sequential Monte Carlo estimators. We reveal that the marginal particle filter is obtained from sequential Monte Carlo by applying Rao-Blackwellization operations, which sacrifices the trajectory information for reduced variance and differentiability. We propose the variational marginal particle filter (VMPF), which is a differentiable and reparameterizable variational filtering objective for SSMs based on an unbiased estimator. We find that VMPF with biased gradients gives tighter bounds than previous objectives, and the unbiased reparameterization gradients are sometimes beneficial.