Semi-Parametric Learning with Bayesian NNs and GPs

We evaluated semi-parametric Gaussian process regression (SPGP) and a novel model-based neural network architecture (BaMbAnn), and compared their modeling accuracy to a series of naive semi-parametric, parametric-only and non-parametric-only regression methods. The comparison has been carried out on three test scenarios, one involving a real test-bed and two involving simulated scenarios, with the most complex scenario targeting the modeling of a simulated 3-link robot’s inverse dynamics model.

Further Material

• [1] Technical Report: Sebastian Riedel, and Freek Stulp. “Comparing Semi-Parametric Model Learning Algorithms for Dynamic Model Estimation in Robotics” [Draft]

---

Technology Stack: Python, Tensorflow, Scikit-Learn, C++, Limbo

Work Affiliation: German Aerospace Center (DLR), in collaboration with Florian Loeffl