Actively Learning Gaussian Process Dynamics

Abstract

Despite the availability of ever more data enabled through modern sensor and computer technology, it still remains an open problem to learn dynamical systems in a sample-efficient way. We propose active learning strategies that leverage information-theoretical properties arising naturally during Gaussian process regression, while respecting constraints on the sampling process imposed by the system dynamics. Sample points are selected in regions with high uncertainty, leading to exploratory behavior and data-efficient training of the model. All results are finally verified in an extensive numerical benchmark.

References in corpus (2)

• Gaussian Process Optimization in the Bandit Setting: No Regret and Experimental Design
• Near-optimal Nonmyopic Value of Information in Graphical Models

Cited by in corpus (6)

• Probabilistic Robust Linear Quadratic Regulators with Gaussian Processes
• Learning-based attacks in Cyber-Physical Systems: Exploration, Detection, and Control Cost trade-offs
• Learning ODE Models with Qualitative Structure Using Gaussian Processes
• Uncertainty-aware Safe Exploratory Planning using Gaussian Process and Neural Control Contraction Metric
• A Receding Horizon Approach for Simultaneous Active Learning and Control using Gaussian Processes
• Active Model Learning using Informative Trajectories for Improved Closed-Loop Control on Real Robots