Bayesian Formulation

Bayesian formulations provide a powerful framework for quantifying uncertainty in various scientific and engineering problems, aiming to move beyond point estimates to obtain full probability distributions over model parameters. Current research focuses on applying Bayesian methods to diverse areas, including system identification (e.g., using stochastic subspace identification and normalizing flows), denoising complex data (e.g., via graph signal processing and maximum a posteriori estimation), and improving machine learning models (e.g., through variational inference and Bayesian neural networks). This approach enhances model interpretability, robustness, and reliability, leading to more informed decision-making in fields ranging from structural health monitoring to AI explainability.