, exhibiting comprehensive advantages in calibration and efficiency over existing methods. Given its data-driven nature, it provides a general framework for parameter estimation while effectively mitigating the non-uniqueness problem across scientific domains.
Zhu et al. (2026) studied this question.