Structured Inference Networks for Nonlinear State Space Models
This work addresses the challenge of scalable and versatile learning for sequential data in machine learning, though it appears incremental as it builds on existing state space model frameworks.
The authors tackled the problem of learning nonlinear state space models by introducing a unified algorithm that simultaneously learns a compiled inference network and the generative model using structured variational approximations. They demonstrated that this approach results in models with significantly higher held-out likelihood on synthetic and real-world datasets.
Gaussian state space models have been used for decades as generative models of sequential data. They admit an intuitive probabilistic interpretation, have a simple functional form, and enjoy widespread adoption. We introduce a unified algorithm to efficiently learn a broad class of linear and non-linear state space models, including variants where the emission and transition distributions are modeled by deep neural networks. Our learning algorithm simultaneously learns a compiled inference network and the generative model, leveraging a structured variational approximation parameterized by recurrent neural networks to mimic the posterior distribution. We apply the learning algorithm to both synthetic and real-world datasets, demonstrating its scalability and versatility. We find that using the structured approximation to the posterior results in models with significantly higher held-out likelihood.