Variational Gaussian Process State-Space Models

Abstract

State-space models have been successfully used for more than fifty years in different areas of science and engineering. We present a procedure for efficient variational Bayesian learning of nonlinear state-space models based on sparse Gaussian processes. The result of learning is a tractable posterior over nonlinear dynamical systems. In comparison to conventional parametric models, we offer the possibility to straightforwardly trade off model capacity and computational cost whilst avoiding overfitting. Our main algorithm uses a hybrid inference approach combining variational Bayes and sequential Monte Carlo. We also present stochastic variational inference and online learning approaches for fast learning with long time series.

Cited by in corpus (32)

• Approximate Bayesian Smoothing with Unknown Process and Measurement Noise Covariances
• When Gaussian Process Meets Big Data: A Review of Scalable GPs
• Neural Adaptive Sequential Monte Carlo
• Linear dynamical neural population models through nonlinear embeddings
• Stability of Gaussian Process State Space Models
• Learning interpretable continuous-time models of latent stochastic dynamical systems
• Recurrent Gaussian Processes
• An Introduction to Gaussian Process Models
• FedLoc: Federated Learning Framework for Data-Driven Cooperative Localization and Location Data Processing
• Sequential Estimation of Gaussian Process-based Deep State-Space Models
• Streaming Variational Monte Carlo
• Probabilistic Robust Linear Quadratic Regulators with Gaussian Processes
• Safe Interactive Model-Based Learning
• Closed-form Inference and Prediction in Gaussian Process State-Space Models
• Probabilistic Recurrent State-Space Models
• Evaluating the squared-exponential covariance function in Gaussian processes with integral observations
• Computationally Efficient Bayesian Learning of Gaussian Process State Space Models
• Non-Factorised Variational Inference in Dynamical Systems
• Structured Variational Inference in Unstable Gaussian Process State Space Models
• A Data-Driven Approach to Dynamically Adjust Resource Allocation for Compute Clusters
• Variational online learning of neural dynamics
• Decoding of neural data using cohomological feature extraction
• Modeling neural dynamics during speech production using a state space variational autoencoder
• Learning dynamical systems with particle stochastic approximation EM
• A projected nonlinear state-space model for forecasting time series signals
• Learning GPLVM with arbitrary kernels using the unscented transformation
• Learning While Navigating: A Practical System Based on Variational Gaussian Process State-Space Model and Smartphone Sensory Data
• Inferring the Structure of Ordinary Differential Equations
• Continual Learning of Multi-modal Dynamics with External Memory
• Conditionally Independent Multiresolution Gaussian Processes
• Incremental Nonlinear System Identification and Adaptive Particle Filtering Using Gaussian Process
• Prediction with Approximated Gaussian Process Dynamical Models