Nonintrusive Surrogate Modeling Using Sparse Random Features With Applications in Crashworthiness Analysis

Efficient surrogate modeling is a key requirement for uncertainty quantification in data-driven scenarios. In this work, a novel approach of using sparse random features for surrogate modeling in combination with self-supervised dimensionality reduction is described. The method is compared to other methods on synthetic and real data obtained from crashworthiness analyses. The results underline the applicability of the here described approach, performing at least as well as polynomial chaos expansions and neural networks.