Brief

Researchers have developed a new framework for creating parsimonious stochastic surrogate models for digital twin applications. This method identifies essential variables from observational data by focusing on those that influence the full conditional distribution of a target quantity, rather than just its mean. The approach combines conditional generative modeling with Gaussian-process-based analysis of variance to iteratively refine influential inputs and uncover interpretable structures. This framework can be applied to various domains, including control systems, reinforcement learning, and economic data, yielding surrogate models with performance comparable to those using the full variable set. AI