Actively Learning Gaussian Process Dynamics
This work addresses the challenge of learning dynamical systems efficiently for applications in fields like robotics or control, but it appears incremental as it builds on existing Gaussian process methods.
The paper tackles the problem of sample-efficient learning of dynamical systems by proposing active learning strategies that use information-theoretical properties from Gaussian process regression to select sample points in high-uncertainty regions, verified through extensive numerical benchmarks.
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.