SCIENTIFIC SEMINAR at FoMICS - Antonietta Mira and Ritabrata Dutta, Dec 16 12:15 Room 355

Approximate Bayesian Computation to quantify parameter and model uncertainty with application to systems and differential equations

SCIENTIFIC SEMINAR at FoMICS

16 December 12:15 - 13:00, Room 355, USI

Antonietta Mira and Ritabrata Dutta

InterDisciplinary Institute of Data Science - Università della Svizzera Italiana 

Differential equation (DE) have been used to describe the laws of nature from the early days of modern science. Presently in all fields of science, systems of ordinary, partial or stochastic DEs are used to describe hypothesized models of natural phenomenon. In practice, systems of DEs are solved to simulate the natural phenomenon for predictive purposes, eg. numerical weather prediction. Though these DEs are well explored by domain-specific research, given an observed dataset, statistical inference on the parameters present in the system of DEs is a challenging problem due to the intractability of the likelihood function of the parameters. Here we propose and illustrate the application of Approximate Bayesian Computation (ABC), a likelihood-free inference scheme for estimation and uncertainty quantification of parameters of DEs. We first describe how ABC can be used to quantify parameter and model uncertainty and then apply ABC for inference in systems of DEs used for numerical weather prediction and for modeling the spread of bacterial infection.

For further informations

https://www.ics.usi.ch/index.php/news/237-fomics-winter-school-at-usi-december-15-19-2016

Info: www.idids.usi.ch

USI Press:  www.usi.ch/highlights