Pedro Gonçalves

Wednesday - 17 October 2018

Time: 4.00pm

Ground Floor Seminar Room

25 Howland Street, London, W1T 4JG

Linking mechanistic models with experimental data through simulation-based statistical inference

Bayesian statistical inference provides a principled framework for linking mechanistic models of neural dynamics with empirical measurements. However, for many mechanistic models of interest, and in particular those relying on numerical simulations, statistical inference has not been possible, or has required bespoke and expensive inference algorithms.  

We overcome this limitation by presenting a method for statistical inference on simulation-based models which can be applied in a `black box' manner to a wide range of models in neuroscience. The key idea is to adaptively simulate multiple data-sets and use them to train a probabilistic neural network to perform statistical inference. Once the network is trained, performing statistical inference is very fast, requiring only a single-forward pass through the network.

We explain how our approach can be used to perform parameter-estimation, and illustrate it on a range of applications, including models of ion-channel-dynamics, single- and multi-compartment Hodgkin Huxley models, and models of spiking networks.

Joint work with Jan-Matthis Lueckmann, Marcel Nonnenmacher, Giacomo Bassetto, Kaan Oecal, and Jakob H. Macke.