I speak: 🇬🇧 🇨🇳 🇩🇪 🇫🇷

Thomas Gaskin

Department of Methodology & Data Science Institute
London School of Economics and Political Science
I am an applied mathematician with a background in physics and computational science. I am interested in hybrid neural models, integrating deep learning into classical mechanistic models. I have published on hybrid approaches to modelling complex, multi-agent systems, with applications to human migration, global trade, econometrics, and infectious diseases. I am also interested in network dynamics, and have published on populist dynamics in social networks. I completed my PhD in Mathematics at Cambridge University with a thesis on neural parameter inference, and am currently a member of the Humanet Lab at the LSE, working on human-machine and machine-machine interaction dynamics.

Deep learning
global migration

A novel and detailed view of international migration flows between all countries since 1990, estimated using deep learning on social, economic, and political covariates.

Populism and social media

A multi-lingual analysis of how information flows through social networks, and how populists reframe international news for a domestic audience.

Neural
parameter inference

How can neural networks be used to calibrate parameters of complex, multi-agent systems, e.g. to better model the spread of infection?

Modelling trade
with optimal transport

Uncovering the hidden dynamics of the global food trade using inverse optimal transport.