Parameter estimation

Develop computational methods for biological digital twins, including parameter estimation, uncertainty quantification, Bayesian inference and physics-informed neural networks. Build scalable simulation software and apply it with experimental and clinical collaborators to development and disease models.

…is cryptographically hard even with parameter access. The talk distinguishes sample generation from parameter estimation and studies a resulting memorization–hallucination trade-off.

Decision-making occurring​ during navigation and learning is widely studied in choice behaviors, but less understood in natural and more continuous settings, especially under stress and uncertainty. This process could be investigated in rodent spatial navigation, which has been modeled with place-cell-based models. However, traditional models usually ignore detailed trajectories or kinematics. Here we extended a place cell-based reinforcement learning model to include detailed kinematics and used it to investigate the role of motivational stress. We performed experiments with two strains of mi

Microorganisms can swim in a variety of environments, interacting with chemicals and other proteins in the fluid. In this talk, we will highlight recent computational methods and results for swimming efficiency and hydrodynamic interactions of swimmers in different fluid environments. Sperm are modeled via a centerline representation where forces are solved for using elastic rod theory. The method of regularized Stokeslets is used to solve the fluid-structure interaction where emergent swimming speeds can be compared to asymptotic analysis. In the case of fluids with extra proteins or cells th