Computer Science (PhD)

"Fast and accurate methods of inference are vital for modelling in the natural sciences. These models tend to contain many free parameters, the inferring of which is a signicant challenge [1]. In particular at multiple scales in neuroscience, models may consist of a stochastic dynamical system which does not admit an analytical solution. Running forward simulations for these systems can be challenging and performing Bayesian inference on them even more so. In this case performing approximate inference is essential to fit the model to observed
data."