Yubo’s work was accepted to NeurIPS 2024 for establishing the identifiability of a hybrid neural-physical deep generative model via a novel use of meta-learning formulation.