PI: Dr. Muhammad Fuady Emzir

This project applies information geometry, which models probability distributions as Riemannian manifolds, to develop efficient continuous-discrete projection filters for parameter estimation in stochastic dynamical systems governed by stochastic differential equations (SDEs) with discrete-time measurements. By employing Bayesian methods such as Markov chain Monte Carlo (MCMC) sampling and variational inference, the approach aims to be more scalable, accurate, and computationally efficient than existing methods. This work seeks to enhance the understanding and interpretation of dynamical systems across fields like engineering, science, and economics.