Quantifying Uncertainty in Symmetric Particle-Based Models for Statistical Inference
Antonin Della Noce (Institut Gustave Roussy)
Abstract: Particle-based or Individual-Based Models (IBMs), initially developed for kinetic gas theory, have found applications across various scientific disciplines, including computational biology, for the purpose of explaining emergent macroscopic phenomena from microscopic interactions. In many applications, some parameters of the system need to be inferred from observation data carrying very partial information on the underlying population / particle assembly This presentation is divided in two parts. The initial part addresses the issue of parameter inference in scenarios where observational data provide limited information about the underlying population or particle assembly. It will discuss the propagation of this partial system knowledge into uncertainties associated with parameter values. The subsequent part focuses on the evaluation of the consistency of mean-field approximations, specifically within the framework of a model representing plant populations in competition for light, which is partially observed.
Bio: Antonin Della Noce obtained his Ph.D. in Applied Mathematics from the Laboratory of Mathematics and Computer Science for Complex Systems (MICS) at Université Paris-Saclay. His doctoral research focused on population dynamical systems. Following his Ph.D., Antonin collaborated with Institut Gustave Roussy to conduct biostatistical research aimed at predicting breast cancer toxicities through the use of high-throughput proteomics. Additionally, he worked with Hôpital Bichat on developing screening strategies for sequencing patients suspected of having connective tissue disorders.