Thomas Uriot

I joined the ACT in June 2019 as a Young Graduate Trainee. I am a Statistician by training, with a BSc in Mathematics from Edinburgh and MSc in Statistical Science from Oxford. Like many of us, and with p-values and rigorous statistical testing falling out of fashion, I have turned towards machine learning.

My research interests range from multiple instance regression (MIR), NLP, and using interpretability techniques in neural networks combined with ideas from Neuroevolution.

At the ACT, I have spent most of my time investigating the potential of mixing deep learning (gradient-based training) with ideas from the field of evolutionary computing, in particular genetic algorithms. This work led to a publication at GECCO 2020.

I am also working on the analysis of one of our Kelvins competition (Collision Avoidance Challenge) alongside the top ranked team and the Space Debris Office at ESA.

In my spare time, I enjoy conducting statistical analysis of NBA games and players, doing some algorithmic trading and watching sport.