On active learning for Gaussian process-based global sensitivity analysis

Download PDF
Open Webpage

Mohit Singh Chauhan, Mariel Ojeda-Tuz, Ryan Catarelli, Kurtis R. Gurley, Dimitrios Tsapetis, Michael D. Shields

Published: 01 Jan 2024, Last Modified: 02 Sept 2025Reliab. Eng. Syst. Saf. 2024EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Strategies for active learning for global sensitivity analysis (GSA) are compared.•Active learning for GSA is difficult due to a ratio of errors that arises.•A new strategy that learns from uncertainties in the main effect is proposed.•The new strategy is compared with existing strategies for a set of test functions.•The new strategy is applied for GSA on Boundary Layer wind tunnel experiments.