Bayesian deep learning to study non-Gaussianity, correlations, and change-points in cell-driven transport (B10*)

• Subject Area: Mathematics
• Term: since 2026
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 318763901

Colloidal particles transported by a “living carpet” of amoeboid cells exhibit strongly nonGaussian displacement distributions, appearing to converge towards an exponential (Laplace) distribution in the long-time limit. Concurrently, the mean-squared displacement is superdiffusive at shorter times and normal-diffusive at longer times. In this project, we will garner new data covering different cell densities and monitoring the dynamics of cell-cargo contacts. We will develop a theoretical description based on stochastic fractional Laplace motion, aided by Bayesian Deep Learning strategies and extend it to analyse change-points in the particle motion.

• DFG Programme: Collaborative Research Centres
• Subproject of: SFB 1294: Data Assimilation – The Seamless Integration of Data and Models
• Applicant Institution: Universität Potsdam
• Project Heads: Professor Dr. Carsten Beta; Professor Dr. Ralf Metzler