Active Learning for Approximation of Expensive Functions with Normal Distributed Output Uncertainty
This work addresses the challenge of efficient function approximation in domains like engineering or science where evaluations are costly, but it appears incremental as it builds on an existing method for deterministic cases.
The paper tackles the problem of approximating expensive black-box functions with output uncertainty by extending the FLOLA-Voronoi active learning method to handle normally distributed uncertainties, resulting in an algorithm that emphasizes exploration to improve model information.
When approximating a black-box function, sampling with active learning focussing on regions with non-linear responses tends to improve accuracy. We present the FLOLA-Voronoi method introduced previously for deterministic responses, and theoretically derive the impact of output uncertainty. The algorithm automatically puts more emphasis on exploration to provide more information to the models.