Hybrid Bayesian Network

Hybrid Bayesian networks (HBNs) combine discrete and continuous variables within a probabilistic framework to model complex systems, aiming to improve the accuracy and efficiency of probabilistic inference. Current research focuses on developing algorithms to handle the computational challenges posed by large, complex HBNs, such as employing factorization techniques to reduce model size and improve inference speed, and incorporating interventional data to enhance causal discovery. These advancements are proving valuable in diverse applications, including risk assessment in medical devices, battery health management in drones, and bioprocess modeling, where they offer improved predictive capabilities and uncertainty quantification.