Probabilistic Recurrent State-Space Models
This work addresses the training challenges of probabilistic state-space models for time series analysis, which is incremental as it builds on existing methods with a new variational approach.
The authors tackled the difficulty of training fully probabilistic state-space models by proposing a novel formulation and scalable training algorithm using doubly stochastic variational inference and Gaussian processes, achieving effectiveness on real-world benchmark datasets and demonstrating scalability on high-dimensional problems.
State-space models (SSMs) are a highly expressive model class for learning patterns in time series data and for system identification. Deterministic versions of SSMs (e.g. LSTMs) proved extremely successful in modeling complex time series data. Fully probabilistic SSMs, however, are often found hard to train, even for smaller problems. To overcome this limitation, we propose a novel model formulation and a scalable training algorithm based on doubly stochastic variational inference and Gaussian processes. In contrast to existing work, the proposed variational approximation allows one to fully capture the latent state temporal correlations. These correlations are the key to robust training. The effectiveness of the proposed PR-SSM is evaluated on a set of real-world benchmark datasets in comparison to state-of-the-art probabilistic model learning methods. Scalability and robustness are demonstrated on a high dimensional problem.