dr Agnieszka Borowska, University of Glasgow, Efficient statistical methods for computationally challenging problems.

Research in physics often starts from a mechanistic model 
and then aims to understand how different parameter regimes are related 
to certain features in the data. This type of research focuses on the 
model and views features in the data as emergent properties. Statistics 
follows the opposite direction: given the data, the aim is to find, or 
"infer", the parameters of the model that are most consistent with the 
observations. In this talk I will focus on three different applications 
in which statistical inference is computationally challenging for 
various reasons and I will show how state-of-the-arr methods from 
computational statistics can come to our rescue. First, I will discuss 
optimising  physiological parameters of a bio-mechanical model of the 
left ventricle for which standard gradient-based optimisation schemes 
are prohibitively time consuming. Second, I will talk about a stochastic 
system describing cell movement with the outputs of the associated 
simulator being very high-dimensional. Third, I will speak about risk 
evaluation, or estimating the probability of a rare event, for which 
simulating from the model directly typically does not lead to improved 
insights on the event in question.