A Bayesian target vector optimization method is presented that enables the fast reconstruction of model parameters from measurements. It combines the advantages of conventional Bayesian optimization with specialized curve fitting algorithms such as the Levenberg-Marquardt method. The method is implemented in JCMsuite's Analysis and Optimization Toolkit.
M. Plock, et al. Recent advances in Bayesian optimization with applications to parameter reconstruction in optical nano-metrology. Proc. SPIE 11783,117830J (2021).
DOI: 10.1117/12.2592266