Publications
Research at Latent Dynamics
Our work sits at the intersection of Bayesian inference, computational physics, and machine learning — producing methods that are both theoretically grounded and practically deployable.
Bayesian inference provides a principled framework for integrating heterogeneous sources of information to estimate latent states under conditions of high noise, limited high-quality data, and partial system observability.
Ensemble convergence of the constrained Bayesian inference method.
Ensemble solution uniqueness under constraint enforcement.
Our unified Bayesian inference and computational framework integrates empirical data, domain-specific knowledge, and multi-physics simulations to infer latent states and predict long-term outcomes in complex systems, enabling applications in disease risk assessment and financial forecasting.
